feat: UUID rate semver FPF-14284

app/build.gradle

@@ -32,7 +32,7 @@ dependencies {
     implementation group: 'org.apache.commons', name: 'commons-text', version: '1.8'
     implementation group: 'org.apache.commons', name: 'commons-math3', version: '3.6.1'
 
-    implementation group: 'com.telefonica.baikal', name: 'spark-commons_2.12', version: '2.2.0'
+    implementation group: 'com.telefonica.baikal', name: 'spark-commons_2.12', version: '55e808c6ddee0d8620bbe381459afe3428f0bec0'
 
     // XXX: custom lib with performance improvements: https://github.com/Telefonica/libphonenumber-1
     implementation group: 'com.googlecode.libphonenumber', name: 'libphonenumber', version: '8.12.25-4p'

app/src/main/java/io/confluent/avro/random/generator/LogicalTypeGenerator.java

@@ -293,12 +293,15 @@ public String random(String logicalType, Map propertiesProp) {
       case "feature-collection":
         int numFeatures = 10;
         ArrayList<Feature> features = new ArrayList<>(numFeatures);
-
+        Feature feat = null;
         for (int i = 0; i < numFeatures; i++) {
           features.add(new Feature(new Geometry(null, null)));
         }
-
         return new FeatureCollection(features).toString();
+      case "uuid-string":
+        return UUID.randomUUID().toString();
+      case "semver":
+        return faker.regexify("(0|[1-9]\\d*)\\.(0|[1-9]\\d*)\\.(0|[1-9]\\d*)(-((0|[1-9]\\d*|\\d*[a-zA-Z-][0-9a-zA-Z-]*)(\\.(0|[1-9]\\d*|\\d*[a-zA-Z-][0-9a-zA-Z-]*))*))?(\\+([0-9a-zA-Z-]+(\\.[0-9a-zA-Z-]+)*))?");
       default:
         throw new IllegalArgumentException("Unsupported logical type: " + logicalType);
     }
@@ -315,4 +318,12 @@ public Double randomDouble(String logicalType, Map propertiesProp) {
     }
   }
 
+  public Float randomFloat(String logicalType, Map propertiesProp) {
+    switch (logicalType) {
+      case "rate":
+        return random.nextFloat();
+      default:
+        throw new IllegalArgumentException("Unsupported logical type: " + logicalType);
+    }
+  }
 }

app/src/main/java/io/confluent/avro/random/generator/geojson/Geometry.java

@@ -2,6 +2,7 @@
 
 import io.confluent.avro.random.generator.geojson.crs.CRS;
 import io.confluent.avro.random.generator.geojson.crs.EPSG4326;
+import com.telefonica.baikal.utils.Validations;
 
 import java.util.Arrays;
 import java.util.List;
@@ -74,27 +75,44 @@ private String randomCoordinates() {
                         )
                 ).toString();
             case MultiPolygon:
-                String first1 = new Point(null, null, crs).toString();
-                String first2 = new Point(null, null, crs).toString();
-
-                return Arrays.asList(
-                        List.of(
+                int i = 0;
+                String result = null;
+                while (i < 10000) {
+                    String first1 = new Point(null, null, crs).toString();
+                    String first2 = new Point(null, null, crs).toString();
+                    Point point1 = new Point(null, null, crs);
+                    Point point2 = new Point(point1.latitude, null, crs);
+                    Point point3 = new Point(null, point2.longitude, crs);
+                    Point point4 = new Point(null, null, crs);
+                    Point point5 = new Point(point4.latitude, null, crs);
+                    Point point6 = new Point(null, point5.longitude, crs);
+                    /* NOTE: The polygon is formed by 5 points due to geojson library */
+                    result = Arrays.asList(
+                            List.of(
                                 Arrays.asList(
                                         first1,
-                                        new Point(null, null, crs).toString(),
-                                        new Point(null, null, crs).toString(),
+                                        point1.toString(),
+                                        point2.toString(),
+                                        point3.toString(),
                                         first1
-                                )
-                        ),
-                        List.of(
+                                ),
                                 Arrays.asList(
                                         first2,
-                                        new Point(null, null, crs).toString(),
-                                        new Point(null, null, crs).toString(),
+                                        point4.toString(),
+                                        point5.toString(),
+                                        point6.toString(),
                                         first2
                                 )
-                        )
-                ).toString();
+                            )
+                    ).toString();
+
+                    String geojson = "{\"type\": \"" + geometryType + "\", \"coordinates\": " + result + "}";
+                    if (Validations.isValidGeometry(geojson)) {
+                        return result;
+                    }
+                    i++;
+                }
+                throw new RuntimeException("Could not generate a valid MultiPolygon after 1000 attempts.");
             default:
                 return null;
         }

app/src/main/java/io/confluent/avro/random/generator/geojson/crs/CRS.java

@@ -22,12 +22,12 @@ protected void setBounds(
 
     public double randomLatitude() {
         Random r = new Random();
-        return bottomBound + (upperBound - bottomBound) * r.nextDouble();
+        return -90 + r.nextInt(180)  + r.nextDouble();
     }
 
     public double randomLongitude() {
         Random r = new Random();
-        return leftBound + (rightBound - leftBound) * r.nextDouble();
+        return -180 + r.nextInt(360) + r.nextDouble();
     }
 
     public abstract String toString();

app/src/test/java/io/confluent/avro/random/generator/LogicalTypeGeneratorTest.java

@@ -281,6 +281,7 @@ public void shouldCreateValidLongitude() {
   @Test
   public void shouldCreateValidLGeometry() {
     GenericRecord record = generateRecordWithSchema("test-schemas/logical-types/geometry.json");
+
     String field = "geometry";
     assertNotNull(record.get(field));
     String value = record.get(field).toString();
@@ -291,6 +292,7 @@ public void shouldCreateValidLGeometry() {
   @Test
   public void shouldCreateValidLFeature() {
     GenericRecord record = generateRecordWithSchema("test-schemas/logical-types/feature.json");
+
     String field = "feature";
     assertNotNull(record.get(field));
     String value = record.get(field).toString();
@@ -301,11 +303,44 @@ public void shouldCreateValidLFeature() {
   @Test
   public void shouldCreateValidLFeatureCollection() {
     GenericRecord record = generateRecordWithSchema("test-schemas/logical-types/feature-collection.json");
+
     String field = "featureCollection";
     assertNotNull(record.get(field));
     String value = record.get(field).toString();
     System.out.println("Generated value is: " + value);
     assertTrue("Invalid feature-collection: " + value, Validations.isValidFeatureCollection(value));
   }
 
+  @Test
+  public void shouldCreateValidUUID() {
+    GenericRecord record = generateRecordWithSchema("test-schemas/logical-types/uuid-string.json");
+
+    String field = "uuid";
+    assertNotNull(record.get(field));
+    String value = record.get(field).toString();
+    System.out.println("Generated value is: " + value);
+    assertTrue("Invalid uuid: " + value, Validations.isValidUUID(value));
+  }
+
+  @Test
+  public void shouldCreateValidRate() {
+    GenericRecord record = generateRecordWithSchema("test-schemas/logical-types/rate.json");
+
+    String field = "rate";
+    assertNotNull(record.get(field));
+    Float value = (float)record.get(field);
+    System.out.println("Generated value is: " + value);
+    assertTrue("Invalid rate: " + value, Validations.isValidRate(value));
+  }
+
+  @Test
+  public void shouldCreateValidSemVer() {
+    GenericRecord record = generateRecordWithSchema("test-schemas/logical-types/semver.json");
+
+    String field = "semver";
+    assertNotNull(record.get(field));
+    String value = record.get(field).toString();
+    System.out.println("Generated value is: " + value);
+    assertTrue("Invalid semver: " + value, Validations.isValidSemVer(value));
+  }
 }

app/src/test/java/io/confluent/avro/random/generator/geojson/FeatureCollectionTest.java

@@ -24,10 +24,12 @@ public void shouldStringifyCorrectly() {
         FeatureCollection featureCollection = createFeatureCollection();
         assertNotNull(featureCollection.toString());
         Gson gson = new Gson();
-
+        String fcollection = featureCollection.toString();
         try {
-            gson.fromJson(featureCollection.toString(), JsonObject.class);
-        } catch (JsonSyntaxException exception) {
+            gson.fromJson(fcollection, JsonObject.class);
+        } catch (Exception exception) {
+
+            exception.printStackTrace();
             fail();
         }
 

app/src/test/resources/test-schemas/logical-types/rate.json

@@ -0,0 +1,14 @@
+{ "type": "record",
+  "name": "logicals",
+  "namespace": "io.confluent.avro.random.generator",
+  "fields":
+  [
+    {
+      "name": "rate",
+      "type": {
+        "type": "float",
+        "logicalType": "rate"
+      }
+    }
+  ]
+}

app/src/test/resources/test-schemas/logical-types/semver.json

@@ -0,0 +1,14 @@
+{ "type": "record",
+  "name": "logicals",
+  "namespace": "io.confluent.avro.random.generator",
+  "fields":
+  [
+    {
+      "name": "semver",
+      "type": {
+        "type": "string",
+        "logicalType": "semver"
+      }
+    }
+  ]
+}

app/src/test/resources/test-schemas/logical-types/uuid-string.json

@@ -0,0 +1,14 @@
+{ "type": "record",
+  "name": "logicals",
+  "namespace": "io.confluent.avro.random.generator",
+  "fields":
+  [
+    {
+      "name": "uuid",
+      "type": {
+        "type": "string",
+        "logicalType": "uuid-string"
+      }
+    }
+  ]
+}

samples/datasets/Full_LogicalType_Dataset_With_Complex_Types/v1.0.0/Full_LogicalType_Dataset_With_Complex_Types_1.0.0.avsc

@@ -21,7 +21,14 @@
     {"name":"IPV6", "type":{"type":"string", "logicalType":"ipv6", "x-fp-data-protection": "pseudonymize"}, "doc":"Internet Protocol Address v6"},
     {"name":"LATITUDE", "type":{"type":"double", "logicalType":"latitude"}, "doc":"Latitude"},
     {"name":"LONGITUDE", "type":{"type":"double", "logicalType":"longitude"}, "doc":"Longitude"},
-    {"name": "SUB_RECORD", "type":{"name":"SUB_RECORD", "type": "record", "fields": [
+    {"name":"DECIMAL", "type": {"type": "string", "logicalType": "decimal-string"}, "doc": "A string representation of a decimal numeral"},
+    {"name":"GEOMETRY", "aliases": ["geometry"], "type": {"type": "string", "logicalType": "geometry"}, "doc": "A Geometry object represents points, curves, and surfaces in coordinate space"},
+    {"name":"FEATURE", "aliases": ["feature"], "type": {"type": "string", "logicalType": "feature"}, "doc": "A Feature object represents a spatially bounded thing"},
+    {"name":"FEATURE_COLLECTION", "aliases": ["feature_collection"], "type": {"type": "string", "logicalType": "feature-collection"}, "doc": "List of features"},
+    {"name":"UUID", "aliases": ["uuid-string"], "type": {"type": "string", "logicalType": "uuid-string"}, "doc": "A string representation of a UUID"},
+    {"name":"RATE", "aliases": ["rate"], "type": {"type": "float", "logicalType": "rate"}, "doc": "A float representation of a rate"},
+    {"name":"SEMVER", "aliases": ["semver"], "type": {"type": "string", "logicalType": "semver"}, "doc": "A string representation of a SemVer"},
+    {"name":"SUB_RECORD", "type":{"name":"SUB_RECORD", "type": "record", "fields": [
           {"name":"PHONE_NUMBER", "type": { "type": "string", "logicalType": "phone-number", "x-fp-data-protection": "pseudonymize" }},
           {"name":"IMSI", "type": { "type":"string", "logicalType":"imsi", "x-fp-data-protection": "pseudonymize" }},
           {"name":"DATETIME", "type": { "type":"string", "logicalType":"datetime" }}

samples/datasets/qa_all_logical_types/v1.0.0/qa_all_logical_types_1.0.0.avsc

@@ -188,6 +188,33 @@
         "logicalType": "feature-collection"
       },
       "doc": "List of features"
+    },
+    {
+      "name": "UUID",
+      "aliases": ["uuid-string"],
+      "type": {
+        "type": "string",
+        "logicalType": "uuid-string"
+      },
+      "doc": "A string representation of a UUID"
+    },
+    {
+      "name": "RATE",
+      "aliases": ["rate"],
+      "type": {
+        "type": "float",
+        "logicalType": "rate"
+      },
+      "doc": "A float representation of a rate"
+    },
+    {
+      "name": "SEMVER",
+      "aliases": ["semver"],
+      "type": {
+        "type": "string",
+        "logicalType": "semver"
+      },
+      "doc": "A string representation of a SemVer"
     }
   ]
 }

samples/datasets/qa_all_logical_types_array_map_records/v1.0.0/qa_all_logical_types_array_map_records_1.0.0.avsc

@@ -170,11 +170,14 @@
                 "name" : "FEATURE_COLLECTION",
                 "type" : {
                   "type" : "string",
-                  "logicalType" : "feature_collection"
+                  "logicalType" : "feature-collection"
                 },
                 "doc" : "List of features",
                 "aliases" : [ "feature_collection" ]
-              } ]
+              },
+              {"name":"UUID","aliases":["uuid"],"type":{"type":"string","logicalType":"uuid-string"},"doc":"A string representation of uuid"},
+              {"name":"SEMVER","aliases":["SemVer"],"type":{"type":"string","logicalType":"semver"},"doc":"A string representation of SemVer"},
+              {"name":"RATE","aliases":["rate"],"type":{"type":"float","logicalType":"rate"},"doc":"A float representation of rate"} ]
           }
         }
       }

samples/datasets/qa_all_logical_types_array_records/v1.0.0/qa_all_logical_types_array_records_1.0.0.avsc

@@ -167,11 +167,14 @@
             "name" : "FEATURE_COLLECTION",
             "type" : {
               "type" : "string",
-              "logicalType" : "feature_collection"
+              "logicalType" : "feature-collection"
             },
             "doc" : "List of features",
             "aliases" : [ "feature_collection" ]
-          } ]
+          },
+          {"name":"UUID","aliases":["uuid"],"type":{"type":"string","logicalType":"uuid-string"},"doc":"A string representation of uuid"},
+          {"name":"SEMVER","aliases":["SemVer"],"type":{"type":"string","logicalType":"semver"},"doc":"A string representation of SemVer"},
+          {"name":"RATE","aliases":["rate"],"type":{"type":"float","logicalType":"rate"},"doc":"A float representation of rate"} ]
         }
       }
     } ],

samples/datasets/qa_all_logical_types_map_records/v1.0.0/qa_all_logical_types_map_records_1.0.0.avsc

@@ -168,11 +168,14 @@
             "name" : "FEATURE_COLLECTION",
             "type" : {
               "type" : "string",
-              "logicalType" : "feature_collection"
+              "logicalType" : "feature-collection"
             },
             "doc" : "List of features",
             "aliases" : [ "feature_collection" ]
-          } ]
+          },
+          {"name":"UUID","aliases":["uuid"],"type":{"type":"string","logicalType":"uuid-string"},"doc":"A string representation of uuid"},
+          {"name":"SEMVER","aliases":["SemVer"],"type":{"type":"string","logicalType":"semver"},"doc":"A string representation of SemVer"},
+          {"name":"RATE","aliases":["rate"],"type":{"type":"float","logicalType":"rate"},"doc":"A float representation of rate"} ]
         }
       }
     } ],

samples/datasets/qa_all_logical_types_nested_records/v1.0.0/qa_all_logical_types_nested_records_1.0.0.avsc

@@ -171,11 +171,14 @@
                 "name" : "FEATURE_COLLECTION",
                 "type" : {
                   "type" : "string",
-                  "logicalType" : "feature_collection"
+                  "logicalType" : "feature-collection"
                 },
                 "doc" : "List of features",
                 "aliases" : [ "feature_collection" ]
-              } ]
+              },
+              {"name":"UUID","aliases":["uuid"],"type":{"type":"string","logicalType":"uuid-string"},"doc":"A string representation of uuid"},
+              {"name":"SEMVER","aliases":["SemVer"],"type":{"type":"string","logicalType":"semver"},"doc":"A string representation of SemVer"},
+              {"name":"RATE","aliases":["rate"],"type":{"type":"float","logicalType":"rate"},"doc":"A float representation of rate"} ]
             }
           } ]
         }