Fixing another problem when the unit comes before in an atomic value #84

lfoppiano · Apr 1, 2019 · 9d74580 · 9d74580
1 parent 9747bd7
commit 9d74580
Show file tree

Hide file tree

Showing 2 changed files with 43 additions and 7 deletions.
diff --git a/src/main/java/org/grobid/core/engines/training/QuantifiedObjectTrainingFormatter.java b/src/main/java/org/grobid/core/engines/training/QuantifiedObjectTrainingFormatter.java
@@ -55,26 +55,45 @@ protected Element trainingExtraction(List<Measurement> measurements, String text
                     endU = unit.getOffsetEnd();
                 }
 
+                boolean addedMeasure = false;
                 int initPos = pos;
 
                 while (pos < text.length()) {
 
                     if (pos == startQ) {
-                        p.appendChild(text.substring(initPos, startQ));
-                        measure.appendChild(text.substring(startQ, endQ));
-                        if (quantifiedObject != null) {
-                            measure.addAttribute(new Attribute("ptr", "#" + quantifiedObjectID));
-//                            markPosition(measure, quantifiedObject, startQ);
+                        if (!addedMeasure) {
+                            if (quantifiedObject != null) {
+                                measure.addAttribute(new Attribute("ptr", "#" + quantifiedObjectID));
+                            }
+                            p.appendChild(text.substring(initPos, startQ));
+                            p.appendChild(measure);
+                            addedMeasure = true;
+                        } else {
+                            measure.appendChild(text.substring(initPos, startQ));
                         }
-                        p.appendChild(measure);
+                        measure.appendChild(text.substring(startQ, endQ));
                         pos = endQ;
                         initPos = pos;
                     }
                     if (pos == startU) {
-                        measure.appendChild(text.substring(initPos, startU));
+                        if (!addedMeasure) {
+                            if (quantifiedObject != null) {
+                                measure.addAttribute(new Attribute("ptr", "#" + quantifiedObjectID));
+                            }
+                            p.appendChild(text.substring(initPos, startU));
+                            p.appendChild(measure);
+                            addedMeasure = true;
+                        } else {
+                            measure.appendChild(text.substring(initPos, startU));
+                        }
                         measure.appendChild(text.substring(startU, endU));
                         pos = endU;
                         initPos = pos;
+
+//                        measure.appendChild(text.substring(initPos, startU));
+//                        measure.appendChild(text.substring(startU, endU));
+//                        pos = endU;
+//                        initPos = pos;
                     }
                     if (pos == startO) {
                         p.appendChild(text.substring(initPos, startO));

diff --git a/src/test/java/org/grobid/core/engines/training/QuantifiedObjectTrainingFormatterTest.java b/src/test/java/org/grobid/core/engines/training/QuantifiedObjectTrainingFormatterTest.java
@@ -421,4 +421,21 @@ public void testTrainingExtraction3() throws Exception {
 
     }
 
+    @Test
+    public void testTriningExtraction4() throws Exception {
+        String text = "Because furosemide-midazolam incompatibility is pH-dependent, the impact of furosemide concentration is predictable. Furosemide in 10 mg/mL of saline is an alkaline solution (pH = 8.77). ";
+
+        LibraryLoader.load();
+        List<Measurement> measurements = QuantityParser.getInstance().process(text);
+//       final Element element = target.trainingExtraction(measurements, text);
+
+        Measurement measurement = new Measurement(UnitUtilities.Measurement_Type.VALUE);
+        measurement.setAtomicQuantity(new Quantity("8.77", new Unit("pH", 175, 177), 180, 184));
+
+        final Element element = target.trainingExtraction(Arrays.asList(measurement), text);
+
+        assertThat(element.toXML(), is("<p xmlns=\"http://www.tei-c.org/ns/1.0\">Because furosemide-midazolam incompatibility is pH-dependent, the impact of furosemide concentration is predictable. Furosemide in 10 mg/mL of saline is an alkaline solution (<measure type=\"value\">pH = 8.77</measure>). </p>"));
+
+    }
+
 }