Improve dependency checker, JSON outout of schema

pkiraly · Dec 11, 2024 · 94033ce · 94033ce
1 parent da4fb10
commit 94033ce
Show file tree

Hide file tree

Showing 8 changed files with 82 additions and 10 deletions.
diff --git a/pom.xml b/pom.xml
@@ -5,7 +5,7 @@
   <groupId>de.gwdg.metadataqa</groupId>
   <artifactId>metadata-qa-api</artifactId>
   <packaging>jar</packaging>
-  <version>0.9.6-SNAPSHOT</version>
+  <version>0.9.7-SNAPSHOT</version>
   <name>Metadata Quality Assurance Framework API</name>
   <description>
     A metadata quality assurance framework. It checks some metrics of 

diff --git a/src/main/java/de/gwdg/metadataqa/api/json/DataElement.java b/src/main/java/de/gwdg/metadataqa/api/json/DataElement.java
@@ -172,6 +172,7 @@ public DataElement setChildren(List<DataElement> children) {
     return this;
   }
 
+  @JsonIgnore
   public boolean isCollection() {
     return collection;
   }

diff --git a/src/main/java/de/gwdg/metadataqa/api/rule/BaseRuleChecker.java b/src/main/java/de/gwdg/metadataqa/api/rule/BaseRuleChecker.java
@@ -91,7 +91,10 @@ else if (outputType.equals(RuleCheckingOutputType.SCORE))
     return header + ":" + getId() + suffix;
   }
 
-  protected void addOutput(FieldCounter<RuleCheckerOutput> results, boolean isNA, boolean allPassed, RuleCheckingOutputType outputType) {
+  protected void addOutput(FieldCounter<RuleCheckerOutput> results,
+                           boolean isNA,
+                           boolean allPassed,
+                           RuleCheckingOutputType outputType) {
     addOutput(results, isNA, allPassed, outputType, null, null);
   }
 

diff --git a/src/main/java/de/gwdg/metadataqa/api/rule/logical/AndChecker.java b/src/main/java/de/gwdg/metadataqa/api/rule/logical/AndChecker.java
@@ -59,7 +59,12 @@ public void update(Selector cache, FieldCounter<RuleCheckerOutput> results, Rule
           MinCountChecker minCountChecker = (MinCountChecker) checker;
           if (!minCountChecker.isEmptyInstancesAllowed() || minCountChecker.getMinCount() > 0)
             allPassed = false;
+        } else if (checker instanceof DependencyChecker) {
+          DependencyChecker dependencyChecker = (DependencyChecker) checker;
+          allPassed = dependencyChecker.getResult(outputType, results);
         }
+        if (!allPassed)
+          break;
       }
     }
     addOutput(results, isNA, allPassed, outputType);

diff --git a/src/main/java/de/gwdg/metadataqa/api/rule/singlefieldchecker/DependencyChecker.java b/src/main/java/de/gwdg/metadataqa/api/rule/singlefieldchecker/DependencyChecker.java
@@ -24,7 +24,9 @@ public DependencyChecker(DataElement field, List<String> dependencies) {
     this(field, field.getLabel(), dependencies, RuleCheckingOutputStatus.FAILED);
   }
 
-  public DependencyChecker(DataElement field, List<String> dependencies, RuleCheckingOutputStatus failedDepencencyStatus) {
+  public DependencyChecker(DataElement field,
+                           List<String> dependencies,
+                           RuleCheckingOutputStatus failedDepencencyStatus) {
     this(field, field.getLabel(), dependencies, failedDepencencyStatus);
   }
 
@@ -42,7 +44,9 @@ public void update(Selector cache, FieldCounter<RuleCheckerOutput> results, Rule
     update(cache, results, outputType, null);
   }
 
-  public void update(Selector cache, FieldCounter<RuleCheckerOutput> localResults, RuleCheckingOutputType outputType,
+  public void update(Selector cache,
+                     FieldCounter<RuleCheckerOutput> localResults,
+                     RuleCheckingOutputType outputType,
                      FieldCounter<RuleCheckerOutput> globalResults) {
     if (isDebug())
       LOGGER.info(this.getClass().getSimpleName() + " " + this.id);
@@ -82,4 +86,31 @@ public void update(Selector cache, FieldCounter<RuleCheckerOutput> localResults,
     if (isDebug())
       LOGGER.info(this.getClass().getSimpleName() + " " + this.id + ") result: " + RuleCheckingOutputStatus.create(isNA, allPassed));
   }
+
+  public List<String> getDependencies() {
+    return dependencies;
+  }
+
+  public boolean getResult(RuleCheckingOutputType outputType,
+                           FieldCounter<RuleCheckerOutput> globalResults) {
+    boolean allPassed = true;
+    for (String ruleId : dependencies) {
+      String keyEnd = outputType.equals(RuleCheckingOutputType.BOTH) ? ruleId + ":status" : ruleId;
+      boolean found = false;
+      for (Map.Entry<String, RuleCheckerOutput> entry : globalResults.getMap().entrySet()) {
+        if (entry.getKey().endsWith(keyEnd)) {
+          found = true;
+          if (entry.getValue().getStatus().equals(RuleCheckingOutputStatus.FAILED)) {
+            allPassed = false;
+            break;
+          }
+        }
+      }
+      if (!found) {
+        allPassed = false;
+        break;
+      }
+    }
+    return allPassed;
+  }
 }
diff --git a/src/test/java/de/gwdg/metadataqa/api/cli/VersionTest.java b/src/test/java/de/gwdg/metadataqa/api/cli/VersionTest.java
@@ -6,7 +6,7 @@
 
 public class VersionTest {
 
-  private final String EXPECTED_VERSION = "0.9.6-SNAPSHOT";
+  private final String EXPECTED_VERSION = "0.9.7-SNAPSHOT";
 
   @Test
   public void getVersion() {

diff --git a/src/test/java/de/gwdg/metadataqa/api/rule/logical/AndCheckerTest.java b/src/test/java/de/gwdg/metadataqa/api/rule/logical/AndCheckerTest.java
@@ -101,4 +101,36 @@ public void failure() {
 
     assertEquals(RuleCheckingOutputStatus.FAILED, fieldCounter.get(andChecker.getHeader(RuleCheckingOutputType.STATUS)).getStatus());
   }
+
+  @Test
+  public void withDependency() {
+    schema.getPathByLabel("name")
+      .setRule(Arrays.asList(
+        new Rule().withAnd(Arrays.asList(
+            new Rule().withMinCount(1),
+            new Rule().withMaxCount(1)))
+          .withNaScore(-1)
+          .withId("Q1"),
+        new Rule().withAnd(Arrays.asList(
+          new Rule().withDependencies(List.of("Q1")),
+          new Rule().withHasValue("a")))
+          .withId("Q2")
+      ));
+
+    cache = (CsvSelector) SelectorFactory.getInstance(schema.getFormat(), ",b,a");
+    cache.setCsvReader(new CsvReader().setHeader(((CsvAwareSchema) schema).getHeader()));
+
+    FieldCounter<RuleCheckerOutput> fieldCounter = new FieldCounter<>();
+    for (RuleChecker checker : schema.getRuleCheckers()) {
+      checker.update(cache, fieldCounter, RuleCheckingOutputType.STATUS);
+    }
+    System.err.println(fieldCounter);
+    // FieldCounter{fieldMap={
+    //   name:and:name:minCount:name:maxCount:Q1:status=0,
+    //   name:and:name:minCount:name:maxCount:Q1:score=0,
+    //   name:and:name:dependency:name:hasValue:Q2:status=0,
+    //   name:and:name:dependency:name:hasValue:Q2:score=0
+    // }}
+
+  }
 }
diff --git a/src/test/java/de/gwdg/metadataqa/api/schema/BaseSchemaTest.java b/src/test/java/de/gwdg/metadataqa/api/schema/BaseSchemaTest.java
@@ -924,17 +924,17 @@ public void toJson() {
     assertTrue(schemaString.contains("\"name\":\"author\""));
     assertTrue(schemaString.contains("\"path\":\"author\""));
     assertTrue(schemaString.contains("\"categories\":[]"));
-    // assertTrue(schemaString.contains("\"children\":[]"));
-    assertTrue(schemaString.contains("\"collection\":false"));
+    assertFalse(schemaString.contains("\"children\":[]"));
+    assertFalse(schemaString.contains("\"collection\":false"));
     assertTrue(schemaString.contains("\"hasValue\":\"a\""));
     assertTrue(schemaString.contains("\"hidden\":false"));
     assertTrue(schemaString.contains("\"skip\":false"));
     assertTrue(schemaString.contains("\"debug\":false"));
     assertTrue(schemaString.contains("\"allowEmptyInstances\":false"));
-    // assertTrue(schemaString.contains("\"absolutePath\":\"author\""));
-    // assertTrue(schemaString.contains("\"active\":true"));
+    assertFalse(schemaString.contains("\"absolutePath\":\"author\""));
+    assertFalse(schemaString.contains("\"active\":true"));
     assertTrue(schemaString.contains("\"extractable\":false"));
-    // assertTrue(schemaString.contains("\"mandatory\":false"));
+    assertFalse(schemaString.contains("\"mandatory\":false"));
     assertTrue(schemaString.contains("\"rules\":["));
     assertTrue(schemaString.contains("\"groups\":[]"));
     assertTrue(schemaString.contains("\"categories\":[]"));