JSON Schema to Avro: implement record writing

Signed-off-by: Francis Galiegue <[email protected]>

Author: fge

src/main/java/com/github/fge/jsonschema2avro/AvroWriterProcessor.java

@@ -14,6 +14,7 @@
 import com.github.fge.jsonschema2avro.writers.ArrayWriter;
 import com.github.fge.jsonschema2avro.writers.EnumWriter;
 import com.github.fge.jsonschema2avro.writers.MapWriter;
+import com.github.fge.jsonschema2avro.writers.RecordWriter;
 import com.github.fge.jsonschema2avro.writers.SimpleTypeWriter;
 import org.apache.avro.Schema;
 
@@ -33,6 +34,7 @@ public AvroWriterProcessor()
             .when(array()).then(ArrayWriter.getInstance())
             .when(map()).then(MapWriter.getInstance())
             .when(isEnum()).then(new EnumWriter())
+            .when(record()).then(new RecordWriter())
             .getProcessor();
     }
 

src/main/java/com/github/fge/jsonschema2avro/predicates/AvroPredicates.java

@@ -131,6 +131,37 @@ public boolean apply(final AvroPayload input)
         };
     }
 
+    public static Predicate<AvroPayload> record()
+    {
+        return new Predicate<AvroPayload>()
+        {
+            @Override
+            public boolean apply(final AvroPayload input)
+            {
+                final JsonNode node = schemaNode(input);
+                final NodeType type = getType(node);
+                if (NodeType.OBJECT != type)
+                    return false;
+
+                if (node.path("additionalProperties").asBoolean(true))
+                    return false;
+
+                if (node.has("patternProperties"))
+                    return false;
+
+                final JsonNode properties = node.path("properties");
+                if (!properties.isObject())
+                    return false;
+
+                for (final String s: Sets.newHashSet(properties.fieldNames()))
+                    if (!isValidAvroName(s))
+                        return false;
+
+                return true;
+            }
+        };
+    }
+
     private static JsonNode schemaNode(final AvroPayload payload)
     {
         return payload.getTree().getNode();

src/main/java/com/github/fge/jsonschema2avro/writers/RecordWriter.java

@@ -0,0 +1,49 @@
+package com.github.fge.jsonschema2avro.writers;
+
+import com.github.fge.jsonschema.exceptions.ProcessingException;
+import com.github.fge.jsonschema.jsonpointer.JsonPointer;
+import com.github.fge.jsonschema.processors.data.SchemaHolder;
+import com.github.fge.jsonschema.report.ProcessingReport;
+import com.github.fge.jsonschema.tree.SchemaTree;
+import com.github.fge.jsonschema2avro.AvroWriterProcessor;
+import com.google.common.collect.Lists;
+import org.apache.avro.Schema;
+
+import java.util.Collections;
+import java.util.List;
+
+public final class RecordWriter
+    extends NamedTypeWriter
+{
+    public RecordWriter()
+    {
+        super("record");
+    }
+
+    @Override
+    protected Schema generate(final AvroWriterProcessor writer,
+        final ProcessingReport report, final SchemaTree tree)
+        throws ProcessingException
+    {
+        final Schema schema = Schema.createRecord(getName(), null, null, false);
+        final List<String> list
+            = Lists.newArrayList(tree.getNode().get("properties").fieldNames());
+        Collections.sort(list);
+
+        final List<Schema.Field> fields = Lists.newArrayList();
+
+        JsonPointer ptr;
+        SchemaHolder holder;
+        Schema fieldSchema;
+
+        for (final String name: list) {
+            ptr = JsonPointer.of("properties", name);
+            holder = new SchemaHolder(tree.append(ptr));
+            fieldSchema = writer.process(report, holder).getValue();
+            fields.add(new Schema.Field(name, fieldSchema, null, null));
+        }
+
+        schema.setFields(fields);
+        return schema;
+    }
+}

src/test/java/com/github/fge/jsonschema2avro/RecordTest.java

@@ -0,0 +1,13 @@
+package com.github.fge.jsonschema2avro;
+
+import java.io.IOException;
+
+public final class RecordTest
+    extends AvroWriterProcessorTest
+{
+    public RecordTest()
+        throws IOException
+    {
+        super("record");
+    }
+}

src/test/resources/jsonschema2avro/record.json

@@ -0,0 +1,39 @@
+[
+    {
+        "jsonSchema": {
+            "type": "object",
+            "additionalProperties": false,
+            "properties": {
+                "id": { "type": "integer" },
+                "name": { "type": "string" },
+                "type": { "enum": [ "food", "drink" ] },
+                "tags": {
+                    "type": "array",
+                    "items": { "type": "string" }
+                }
+            }
+        },
+        "avroSchema": {
+            "type": "record",
+            "name": "record0",
+            "fields": [
+                { "name": "id", "type": "long" },
+                { "name": "name", "type": "string" },
+                {
+                    "name": "tags",
+                    "type": {
+                        "type": "array", "items": "string"
+                    }
+                },
+                {
+                    "name": "type",
+                    "type": {
+                        "type": "enum",
+                        "name": "enum0",
+                        "symbols": [ "food", "drink" ]
+                    }
+                }
+            ]
+        }
+    }
+]