SONARPY-2477 Collect data for Databricks notebooks (#2260)

SonarSource · Dec 18, 2024 · 10d865d · 10d865d
1 parent 5b243a7
commit 10d865d
Show file tree

Hide file tree

Showing 7 changed files with 67 additions and 3 deletions.
diff --git a/sonar-python-plugin/src/main/java/org/sonar/plugins/python/IPynbSensor.java b/sonar-python-plugin/src/main/java/org/sonar/plugins/python/IPynbSensor.java
@@ -93,6 +93,7 @@ private void processNotebooksFiles(List<PythonInputFile> pythonFiles, SensorCont
     PythonScanner scanner = new PythonScanner(context, checks, fileLinesContextFactory, noSonarFilter, PythonParser.createIPythonParser(), pythonIndexer);
     scanner.execute(pythonFiles, context);
     sensorTelemetryStorage.updateMetric(TelemetryMetricKey.NOTEBOOK_RECOGNITION_ERROR_KEY, scanner.getRecognitionErrorCount());
+    updateDatabricksTelemetry(scanner);
   }
 
   private List<PythonInputFile> parseNotebooks(List<PythonInputFile> pythonFiles, SensorContext context) {
@@ -135,4 +136,9 @@ private static List<PythonInputFile> getInputFiles(SensorContext context) {
   private static boolean isErrorOnTestFile(PythonInputFile inputFile) {
     return inputFile.wrappedFile().type() == InputFile.Type.TEST;
   }
+
+  private void updateDatabricksTelemetry(PythonScanner scanner) {
+    sensorTelemetryStorage.updateMetric(TelemetryMetricKey.IPYNB_DATABRICKS_FOUND, scanner.getFoundDatabricks());
+  }
+
 }
diff --git a/sonar-python-plugin/src/main/java/org/sonar/plugins/python/PythonScanner.java b/sonar-python-plugin/src/main/java/org/sonar/plugins/python/PythonScanner.java
@@ -30,6 +30,7 @@
 import java.util.Map;
 import java.util.Optional;
 import java.util.Set;
+import java.util.regex.Pattern;
 import javax.annotation.CheckForNull;
 import org.slf4j.Logger;
 import org.slf4j.LoggerFactory;
@@ -78,6 +79,8 @@ public class PythonScanner extends Scanner {
   private final PythonIndexer indexer;
   private final Map<PythonInputFile, Set<PythonCheck>> checksExecutedWithoutParsingByFiles = new HashMap<>();
   private int recognitionErrorCount = 0;
+  private static final Pattern DATABRICKS_MAGIC_COMMAND_PATTERN = Pattern.compile("^\\h*#\\h*(MAGIC|COMMAND).*");
+  private boolean foundDatabricks = false;
 
   public PythonScanner(
     SensorContext context, PythonChecks checks,
@@ -150,6 +153,13 @@ protected void scanFile(PythonInputFile inputFile) throws IOException {
       new SymbolVisitor(context.newSymbolTable().onFile(inputFile.wrappedFile())).visitFileInput(visitorContext.rootTree());
       new PythonHighlighter(context, inputFile).scanFile(visitorContext);
     }
+
+    searchForDataBricks(visitorContext);
+  }
+
+  private void searchForDataBricks(PythonVisitorContext visitorContext) {
+    foundDatabricks |= visitorContext.pythonFile().content().lines().anyMatch(
+      line -> DATABRICKS_MAGIC_COMMAND_PATTERN.matcher(line).matches());
   }
 
   private static PythonTreeMaker getTreeMaker(PythonInputFile inputFile) {
@@ -407,4 +417,8 @@ private static TextRange rangeFromTextSpan(InputFile file, PythonTextEdit python
   public int getRecognitionErrorCount() {
     return recognitionErrorCount;
   }
+
+  public boolean getFoundDatabricks() {
+    return foundDatabricks;
+  }
 }
diff --git a/sonar-python-plugin/src/main/java/org/sonar/plugins/python/PythonSensor.java b/sonar-python-plugin/src/main/java/org/sonar/plugins/python/PythonSensor.java
@@ -131,10 +131,15 @@ public void execute(SensorContext context) {
     TypeShed.setProjectLevelSymbolTable(pythonIndexer.projectLevelSymbolTable());
     PythonScanner scanner = new PythonScanner(context, checks, fileLinesContextFactory, noSonarFilter, PythonParser.create(), pythonIndexer);
     scanner.execute(pythonFiles, context);
+    updateDatabricksTelemetry(scanner);
     sensorTelemetryStorage.send(context);
     durationReport.stop();
   }
 
+  private void updateDatabricksTelemetry(PythonScanner scanner) {
+    sensorTelemetryStorage.updateMetric(TelemetryMetricKey.PYTHON_DATABRICKS_FOUND, scanner.getFoundDatabricks());
+  }
+
   private void updatePythonVersionTelemetry(SensorContext context, String[] pythonVersionParameter) {
     if (context.runtime().getProduct() == SonarProduct.SONARLINT) {
       return;

diff --git a/sonar-python-plugin/src/main/java/org/sonar/plugins/python/SensorTelemetryStorage.java b/sonar-python-plugin/src/main/java/org/sonar/plugins/python/SensorTelemetryStorage.java
@@ -35,12 +35,12 @@ public void send(SensorContext sensorContext) {
       var apiVersion = sensorContext.runtime().getApiVersion();
       if (apiVersion.isGreaterThanOrEqual(Version.create(10, 9))) {
         data.forEach((k, v) -> {
-          LOG.info("Collected metric: {}={}", k, v);
+          LOG.debug("Collected metric: {}={}", k, v);
           sensorContext.addTelemetryProperty(k.key(), v);
         });
 
       } else {
-        LOG.info("Skipping sending metrics because the plugin API version is {}", apiVersion);
+        LOG.debug("Skipping sending metrics because the plugin API version is {}", apiVersion);
       }
     } catch (Exception e) {
       LOG.error("Failed to send metrics", e);

diff --git a/sonar-python-plugin/src/main/java/org/sonar/plugins/python/TelemetryMetricKey.java b/sonar-python-plugin/src/main/java/org/sonar/plugins/python/TelemetryMetricKey.java
@@ -22,7 +22,9 @@ public enum TelemetryMetricKey {
   NOTEBOOK_RECOGNITION_ERROR_KEY("python.notebook.recognition_error"),
   NOTEBOOK_EXCEPTION_KEY("python.notebook.exceptions"),
   PYTHON_VERSION_SET_KEY("python.version.set"),
-  PYTHON_VERSION_KEY("python.version");
+  PYTHON_VERSION_KEY("python.version"),
+  PYTHON_DATABRICKS_FOUND("python.notebook.databricks.python"),
+  IPYNB_DATABRICKS_FOUND("python.notebook.databricks.ipynb");
 
   private final String key;
 

diff --git a/sonar-python-plugin/src/test/java/org/sonar/plugins/python/PythonSensorTest.java b/sonar-python-plugin/src/test/java/org/sonar/plugins/python/PythonSensorTest.java
@@ -1438,6 +1438,36 @@ void send_telemetry_no_version() {
     verify(contextSpy, times(1)).addTelemetryProperty(TelemetryMetricKey.PYTHON_VERSION_SET_KEY.key(), "0");
   }
 
+  @Test
+  void detects_databricks() {
+    activeRules = new ActiveRulesBuilder()
+      .addRule(new NewActiveRule.Builder()
+        .setRuleKey(RuleKey.of(CheckList.REPOSITORY_KEY, "PrintStatementUsage"))
+        .setName("Print Statement Usage")
+        .build())
+      .build();
+
+    inputFile("databricks.py");
+    var spyContext = spy(context);
+    sensor().execute(spyContext);
+    verify(spyContext, times(1)).addTelemetryProperty(TelemetryMetricKey.PYTHON_DATABRICKS_FOUND.key(), "1");
+  }
+
+  @Test
+  void detects_databricks_negative() {
+    activeRules = new ActiveRulesBuilder()
+      .addRule(new NewActiveRule.Builder()
+        .setRuleKey(RuleKey.of(CheckList.REPOSITORY_KEY, "PrintStatementUsage"))
+        .setName("Print Statement Usage")
+        .build())
+      .build();
+
+    inputFile(FILE_1);
+    var spyContext = spy(context);
+    sensor().execute(spyContext);
+    verify(spyContext, times(1)).addTelemetryProperty(TelemetryMetricKey.PYTHON_DATABRICKS_FOUND.key(), "0");
+  }
+
   private com.sonar.sslr.api.Token passToken(URI uri) {
     return com.sonar.sslr.api.Token.builder()
       .setType(PythonKeyword.PASS)

diff --git a/sonar-python-plugin/src/test/resources/org/sonar/plugins/python/sensor/databricks.py b/sonar-python-plugin/src/test/resources/org/sonar/plugins/python/sensor/databricks.py
@@ -0,0 +1,7 @@
+# Databricks notebooks
+# COMMAND ----------
+
+# MAGIC %md
+# MAGIC ## Alter tables
+
+# COMMAND ----------