Update Beam and SmilesParser to be more relaxed on double bond stereo issues. Beam changes

johnmay · johnmay · commit f37f053be4e4 · 2018-12-21T10:02:54.000Z
- allowing strict/relaxed parsing from external API
- better reporting of double bond stereo errors
- ignore double bond stereo errors if in relaxed mode
diff --git a/pom.xml b/pom.xml
@@ -415,12 +415,12 @@
             <dependency>
                 <groupId>uk.ac.ebi.beam</groupId>
                 <artifactId>beam-core</artifactId>
-                <version>1.2</version>
+                <version>1.3.1</version>
             </dependency>
             <dependency>
                 <groupId>uk.ac.ebi.beam</groupId>
                 <artifactId>beam-func</artifactId>
-                <version>1.2</version>
+                <version>1.3.1</version>
             </dependency>
             <dependency>
                 <groupId>com.google.guava</groupId>
diff --git a/storage/smiles/src/main/java/org/openscience/cdk/smiles/BeamToCDK.java b/storage/smiles/src/main/java/org/openscience/cdk/smiles/BeamToCDK.java
@@ -420,11 +420,17 @@ private Edge findDirectionalEdge(Graph g, int u) {
         List<Edge> edges = g.edges(u);
         if (edges.size() == 1)
             return null;
+        Edge first = null;
         for (Edge e : edges) {
             Bond b = e.bond();
-            if (b == Bond.UP || b == Bond.DOWN) return e;
+            if (b == Bond.UP || b == Bond.DOWN) {
+                if (first == null)
+                    first = e;
+                else if (((first.either() == e.either()) == (first.bond() == b)))
+                    return null;
+            }
         }
-        return null;
+        return first;
     }
 
     /**
diff --git a/storage/smiles/src/main/java/org/openscience/cdk/smiles/SmilesParser.java b/storage/smiles/src/main/java/org/openscience/cdk/smiles/SmilesParser.java
@@ -43,6 +43,8 @@
 import org.openscience.cdk.sgroup.SgroupType;
 import org.openscience.cdk.smiles.CxSmilesState.DataSgroup;
 import org.openscience.cdk.smiles.CxSmilesState.PolymerSgroup;
+import org.openscience.cdk.tools.ILoggingTool;
+import org.openscience.cdk.tools.LoggingToolFactory;
 import org.openscience.cdk.tools.manipulator.AtomContainerManipulator;
 import uk.ac.ebi.beam.Graph;
 
@@ -130,6 +132,8 @@
  */
 public final class SmilesParser {
 
+    private ILoggingTool logger = LoggingToolFactory.createLoggingTool(SmilesParser.class);
+
     /**
      * The builder determines which CDK domain objects to create.
      */
@@ -149,6 +153,11 @@ public final class SmilesParser {
      */
     private boolean                  kekulise = true;
 
+    /**
+     * Whether the parser is in strict mode or not.
+     */
+    private boolean                  strict = false;
+
     /**
      * Create a new SMILES parser which will create {@link IAtomContainer}s with
      * the specified builder.
@@ -160,6 +169,16 @@ public SmilesParser(final IChemObjectBuilder builder) {
         this.beamToCDK = new BeamToCDK(builder);
     }
 
+    /**
+     * Sets whether the parser is in strict mode. In non-strict mode (default)
+     * recoverable issues with SMILES are reported as warnings.
+     *
+     * @param strict strict mode true/false.
+     */
+    public void setStrict(boolean strict) {
+        this.strict = strict;
+    }
+
     /**
      * Parse a reaction SMILES.
      *
@@ -239,8 +258,11 @@ public IAtomContainer parseSmiles(String smiles) throws InvalidSmilesException {
 
     private IAtomContainer parseSmiles(String smiles, boolean isRxnPart) throws InvalidSmilesException {
         try {
-            // create the Beam object from the SMILES
-            Graph g = Graph.fromSmiles(smiles);
+            // create the Beam object from parsing the SMILES
+            Set<String> warnings = new HashSet<>();
+            Graph g = Graph.parse(smiles, strict, warnings);
+            for (String warning : warnings)
+              logger.warn(warning);
 
             // convert the Beam object model to the CDK - note exception thrown
             // if a kekule structure could not be assigned.
diff --git a/storage/smiles/src/test/java/org/openscience/cdk/smiles/SmilesParserTest.java b/storage/smiles/src/test/java/org/openscience/cdk/smiles/SmilesParserTest.java
@@ -2663,6 +2663,27 @@ public void notExtendedCis() throws Exception {
         assertFalse(mol.stereoElements().iterator().hasNext());
     }
 
+    @Test
+    public void warnOnDirectionalBonds() throws InvalidSmilesException {
+        SmilesParser smipar = new SmilesParser(SilentChemObjectBuilder.getInstance());
+        IAtomContainer mol = smipar.parseSmiles("C/C=C/1.C/1");
+    }
+
+    @Test(expected = InvalidSmilesException.class)
+    public void failOnDirectionalBondsWhenStrict() throws InvalidSmilesException {
+        SmilesParser smipar = new SmilesParser(SilentChemObjectBuilder.getInstance());
+        smipar.setStrict(true);
+        IAtomContainer mol = smipar.parseSmiles("C/C=C/1.C/1");
+    }
+
+    @Test
+    public void ignoreDoubleBond() throws InvalidSmilesException {
+        SmilesParser smipar = new SmilesParser(SilentChemObjectBuilder.getInstance());
+        IAtomContainer mol = smipar.parseSmiles("C/C=C(/F)/C");
+        assertThat(mol.stereoElements().iterator().hasNext(),
+                   is(false));
+    }
+
     @Test public void extendedTetrahedral7() throws InvalidSmilesException {
         IAtomContainer mol = load("CC=C=C=[C@]=C=C=CC");
         for (IStereoElement se : mol.stereoElements()) {
