SONAR-5031 Provide line count as attribute of InputFile

3 files changed, 53 insertions, 32 deletions

diff --git a/sonar-batch/src/main/java/org/sonar/batch/scan/filesystem/ComponentIndexer.java b/sonar-batch/src/main/java/org/sonar/batch/scan/filesystem/ComponentIndexer.java
index ad0064f565c..45224104444 100644
--- a/sonar-batch/src/main/java/org/sonar/batch/scan/filesystem/ComponentIndexer.java
+++ b/sonar-batch/src/main/java/org/sonar/batch/scan/filesystem/ComponentIndexer.java
@@ -19,9 +19,7 @@
  */
 package org.sonar.batch.scan.filesystem;
 
-import org.sonar.api.scan.filesystem.internal.DefaultInputFile;
-
-import org.sonar.api.scan.filesystem.InputFile;
+import com.google.common.annotations.VisibleForTesting;
 import com.google.common.base.CharMatcher;
 import com.google.common.io.Files;
 import org.apache.commons.lang.StringUtils;
@@ -37,6 +35,8 @@ import org.sonar.api.resources.Languages;
 import org.sonar.api.resources.Project;
 import org.sonar.api.resources.Resource;
 import org.sonar.api.scan.filesystem.FileQuery;
+import org.sonar.api.scan.filesystem.InputFile;
+import org.sonar.api.scan.filesystem.internal.DefaultInputFile;
 import org.sonar.api.utils.SonarException;
 import org.sonar.batch.index.ResourceKeyMigration;
 import org.sonar.batch.scan.language.DefaultModuleLanguages;
@@ -57,9 +57,10 @@ public class ComponentIndexer implements BatchComponent {
   private final Project module;
   private final DefaultModuleLanguages moduleLanguages;
   private final ResourceDao resourceDao;
+  private InputFileCache fileCache;
 
   public ComponentIndexer(Project module, Languages languages, SonarIndex sonarIndex, Settings settings, ResourceKeyMigration migration,
-    DefaultModuleLanguages moduleLanguages, ResourceDao resourceDao) {
+    DefaultModuleLanguages moduleLanguages, ResourceDao resourceDao, InputFileCache fileCache) {
     this.module = module;
     this.languages = languages;
     this.sonarIndex = sonarIndex;
@@ -67,10 +68,11 @@ public class ComponentIndexer implements BatchComponent {
     this.migration = migration;
     this.moduleLanguages = moduleLanguages;
     this.resourceDao = resourceDao;
+    this.fileCache = fileCache;
   }
 
   public void execute(DefaultModuleFileSystem fs) {
-    boolean importSource = settings.getBoolean(CoreProperties.CORE_IMPORT_SOURCES_PROPERTY);
+    boolean shouldImportSource = settings.getBoolean(CoreProperties.CORE_IMPORT_SOURCES_PROPERTY);
     Iterable<InputFile> inputFiles = fs.inputFiles(FileQuery.all());
     migration.migrateIfNeeded(module, inputFiles);
     for (InputFile inputFile : inputFiles) {
@@ -85,21 +87,28 @@ public class ComponentIndexer implements BatchComponent {
       if (sonarFile != null) {
         moduleLanguages.addLanguage(languageKey);
         sonarIndex.index(sonarFile);
-        if (importSource) {
-          importSources(inputFile, sonarFile);
-        }
+        importSources(shouldImportSource, inputFile, sonarFile);
       }
     }
 
     updateModuleLanguage();
   }
 
-  private void importSources(InputFile inputFile, Resource sonarFile) {
+  @VisibleForTesting
+  void importSources(boolean shouldImportSource, InputFile inputFile, Resource sonarFile) {
     try {
+      // TODO this part deserve optimization.
+      // No need to read full content in memory when shouldImportSource=false
+      // We should try to remove BOM and count lines in a single pass
       String source = Files.toString(inputFile.file(), inputFile.encoding());
       // SONAR-3860 Remove BOM character from source
       source = CharMatcher.anyOf("\uFEFF").removeFrom(source);
-      sonarIndex.setSource(sonarFile, source);
+      String[] lines = source.split("(\r)?\n|\r", -1);
+      inputFile.attributes().put(InputFile.ATTRIBUTE_LINE_COUNT, String.valueOf(lines.length));
+      fileCache.put(module.getKey(), inputFile);
+      if (shouldImportSource) {
+        sonarIndex.setSource(sonarFile, source);
+      }
     } catch (Exception e) {
       throw new SonarException("Unable to read and import the source file : '" + inputFile.absolutePath() + "' with the charset : '"
         + inputFile.encoding() + "'.", e);
diff --git a/sonar-batch/src/test/java/org/sonar/batch/scan/filesystem/ComponentIndexerTest.java b/sonar-batch/src/test/java/org/sonar/batch/scan/filesystem/ComponentIndexerTest.java
index c0945f0f570..69c01e7ffab 100644
--- a/sonar-batch/src/test/java/org/sonar/batch/scan/filesystem/ComponentIndexerTest.java
+++ b/sonar-batch/src/test/java/org/sonar/batch/scan/filesystem/ComponentIndexerTest.java
@@ -19,8 +19,6 @@
  */
 package org.sonar.batch.scan.filesystem;
 
-import org.sonar.api.scan.filesystem.InputFile;
-
 import com.google.common.base.Charsets;
 import org.apache.commons.io.FileUtils;
 import org.apache.commons.lang.CharEncoding;
@@ -41,6 +39,7 @@ import org.sonar.api.resources.Project;
 import org.sonar.api.resources.Qualifiers;
 import org.sonar.api.resources.Resource;
 import org.sonar.api.scan.filesystem.FileQuery;
+import org.sonar.api.scan.filesystem.InputFile;
 import org.sonar.api.scan.filesystem.internal.DefaultInputFile;
 import org.sonar.api.scan.filesystem.internal.InputFileBuilder;
 import org.sonar.batch.index.ResourceKeyMigration;
@@ -52,6 +51,7 @@ import java.io.IOException;
 import java.nio.charset.Charset;
 import java.util.Arrays;
 
+import static org.fest.assertions.Assertions.assertThat;
 import static org.mockito.Matchers.argThat;
 import static org.mockito.Matchers.eq;
 import static org.mockito.Mockito.mock;
@@ -96,10 +96,9 @@ public class ComponentIndexerTest {
       newInputFile("src/main/java/foo/bar/Foo.java", "", "foo/bar/Foo.java", "java", false),
       newInputFile("src/main/java2/foo/bar/Foo.java", "", "foo/bar/Foo.java", "java", false),
       newInputFile("src/test/java/foo/bar/FooTest.java", "", "foo/bar/FooTest.java", "java", true)));
-    when(project.getLanguageKey()).thenReturn(Java.KEY);
     Languages languages = new Languages(Java.INSTANCE);
     ComponentIndexer indexer = new ComponentIndexer(project, languages, sonarIndex, settings, mock(ResourceKeyMigration.class), new DefaultModuleLanguages(settings, languages),
-      mock(ResourceDao.class));
+      mock(ResourceDao.class), mock(InputFileCache.class));
     indexer.execute(fs);
 
     verify(sonarIndex).index(JavaFile.create("src/main/java/foo/bar/Foo.java", "foo/bar/Foo.java", false));
@@ -121,11 +120,10 @@ public class ComponentIndexerTest {
       newInputFile("src/foo/bar/Foo.cbl", "", "foo/bar/Foo.cbl", "cobol", false),
       newInputFile("src2/foo/bar/Foo.cbl", "", "foo/bar/Foo.cbl", "cobol", false),
       newInputFile("src/test/foo/bar/FooTest.cbl", "", "foo/bar/FooTest.cbl", "cobol", true)));
-    when(project.getLanguageKey()).thenReturn("cobol");
 
     Languages languages = new Languages(cobolLanguage);
     ComponentIndexer indexer = new ComponentIndexer(project, languages, sonarIndex, settings, mock(ResourceKeyMigration.class), new DefaultModuleLanguages(settings, languages),
-      mock(ResourceDao.class));
+      mock(ResourceDao.class), mock(InputFileCache.class));
     indexer.execute(fs);
 
     verify(sonarIndex).index(org.sonar.api.resources.File.create("/src/foo/bar/Foo.cbl", "foo/bar/Foo.cbl", cobolLanguage, false));
@@ -139,10 +137,9 @@ public class ComponentIndexerTest {
 
     when(fs.inputFiles(FileQuery.all())).thenReturn((Iterable) Arrays.asList(
       newInputFile("src/main/java/foo/bar/Foo.java", "sample code", "foo/bar/Foo.java", "java", false)));
-    when(project.getLanguageKey()).thenReturn(Java.KEY);
     Languages languages = new Languages(Java.INSTANCE);
     ComponentIndexer indexer = new ComponentIndexer(project, languages, sonarIndex, settings, mock(ResourceKeyMigration.class), new DefaultModuleLanguages(settings, languages),
-      mock(ResourceDao.class));
+      mock(ResourceDao.class), mock(InputFileCache.class));
     indexer.execute(fs);
 
     Resource sonarFile = JavaFile.create("src/main/java/foo/bar/Foo.java", "foo/bar/Foo.java", false);
@@ -150,16 +147,6 @@ public class ComponentIndexerTest {
     verify(sonarIndex).setSource(sonarFile, "sample code");
   }
 
-  private DefaultInputFile newInputFile(String path, String content, String sourceRelativePath, String languageKey, boolean unitTest) throws IOException {
-    File file = new File(baseDir, path);
-    FileUtils.write(file, content);
-    return new InputFileBuilder(file, Charsets.UTF_8, path)
-      .attribute(DefaultInputFile.ATTRIBUTE_SOURCE_RELATIVE_PATH, sourceRelativePath)
-      .attribute(InputFile.ATTRIBUTE_LANGUAGE, languageKey)
-      .attribute(InputFile.ATTRIBUTE_TYPE, unitTest ? InputFile.TYPE_TEST : InputFile.TYPE_MAIN)
-      .build();
-  }
-
   @Test
   public void should_use_mac_roman_charset_for_reading_source_files() throws Exception {
     String encoding = "MacRoman";
@@ -192,10 +179,9 @@ public class ComponentIndexerTest {
         .attribute(DefaultInputFile.ATTRIBUTE_SOURCE_RELATIVE_PATH, "foo/bar/Foo.java")
         .attribute(InputFile.ATTRIBUTE_LANGUAGE, "java")
         .build()));
-    when(project.getLanguageKey()).thenReturn(Java.KEY);
     Languages languages = new Languages(Java.INSTANCE);
     ComponentIndexer indexer = new ComponentIndexer(project, languages, sonarIndex, settings, mock(ResourceKeyMigration.class), new DefaultModuleLanguages(settings, languages),
-      mock(ResourceDao.class));
+      mock(ResourceDao.class), mock(InputFileCache.class));
     indexer.execute(fs);
 
     Resource sonarFile = JavaFile.create("src/main/java/foo/bar/Foo.java", "foo/bar/Foo.java", false);
@@ -221,10 +207,9 @@ public class ComponentIndexerTest {
             .attribute(DefaultInputFile.ATTRIBUTE_SOURCE_RELATIVE_PATH, "foo/bar/Foo.java")
             .attribute(InputFile.ATTRIBUTE_LANGUAGE, "java")
             .build()));
-    when(project.getLanguageKey()).thenReturn(Java.KEY);
     Languages languages = new Languages(Java.INSTANCE);
     ComponentIndexer indexer = new ComponentIndexer(project, languages, sonarIndex, settings, mock(ResourceKeyMigration.class), new DefaultModuleLanguages(settings, languages),
-      mock(ResourceDao.class));
+      mock(ResourceDao.class), mock(InputFileCache.class));
     indexer.execute(fs);
 
     Resource sonarFile = JavaFile.create("/src/main/java/foo/bar/Foo.java", "foo/bar/Foo.java", false);
@@ -238,8 +223,30 @@ public class ComponentIndexerTest {
     }));
   }
 
+  @Test
+  public void should_compute_line_count() throws IOException {
+    DefaultInputFile inputFile = newInputFile("src/main/java/foo/bar/Foo.java", "line1\nline2\nline3", "foo/bar/Foo.java", "java", false);
+    when(fs.inputFiles(FileQuery.all())).thenReturn((Iterable) Arrays.asList(inputFile));
+    Languages languages = new Languages(Java.INSTANCE);
+    ComponentIndexer indexer = new ComponentIndexer(project, languages, sonarIndex, settings, mock(ResourceKeyMigration.class), new DefaultModuleLanguages(settings, languages),
+      mock(ResourceDao.class), mock(InputFileCache.class));
+    indexer.execute(fs);
+
+    assertThat(inputFile.attribute(InputFile.ATTRIBUTE_LINE_COUNT)).isEqualTo("3");
+  }
+
   private File getFile(String testFile) {
     return new File("test-resources/org/sonar/batch/phases/FileIndexerTest/encoding/" + testFile);
   }
 
+  private DefaultInputFile newInputFile(String path, String content, String sourceRelativePath, String languageKey, boolean unitTest) throws IOException {
+    File file = new File(baseDir, path);
+    FileUtils.write(file, content);
+    return new InputFileBuilder(file, Charsets.UTF_8, path)
+      .attribute(DefaultInputFile.ATTRIBUTE_SOURCE_RELATIVE_PATH, sourceRelativePath)
+      .attribute(InputFile.ATTRIBUTE_LANGUAGE, languageKey)
+      .attribute(InputFile.ATTRIBUTE_TYPE, unitTest ? InputFile.TYPE_TEST : InputFile.TYPE_MAIN)
+      .build();
+  }
+
 }
diff --git a/sonar-plugin-api/src/main/java/org/sonar/api/scan/filesystem/InputFile.java b/sonar-plugin-api/src/main/java/org/sonar/api/scan/filesystem/InputFile.java
index 91ed1c144c1..e811576ec2e 100644
--- a/sonar-plugin-api/src/main/java/org/sonar/api/scan/filesystem/InputFile.java
+++ b/sonar-plugin-api/src/main/java/org/sonar/api/scan/filesystem/InputFile.java
@@ -34,6 +34,11 @@ public interface InputFile extends Serializable {
   String ATTRIBUTE_LANGUAGE = "LANG";
 
   /**
+   * Number of lines in the file.
+   */
+  String ATTRIBUTE_LINE_COUNT = "LINE_COUNT";
+
+  /**
    * Type of source file. For now only possible values are {@link #TYPE_MAIN} or {@link #TYPE_TEST}
    */
   String ATTRIBUTE_TYPE = "TYPE";