SONAR-5868 Allow issue tracking mechanism to work without full previous sources

21 files changed, 199 insertions, 80 deletions

diff --git a/server/sonar-server/src/main/java/org/sonar/server/db/migrations/v50/FeedFileSources.java b/server/sonar-server/src/main/java/org/sonar/server/db/migrations/v50/FeedFileSources.java
index 3c09e05947f..d7f698f1800 100644
--- a/server/sonar-server/src/main/java/org/sonar/server/db/migrations/v50/FeedFileSources.java
+++ b/server/sonar-server/src/main/java/org/sonar/server/db/migrations/v50/FeedFileSources.java
@@ -21,14 +21,15 @@ package org.sonar.server.db.migrations.v50;
 
 import org.sonar.api.utils.System2;
 import org.sonar.core.persistence.Database;
-import org.sonar.server.db.migrations.*;
+import org.sonar.server.db.migrations.BaseDataChange;
+import org.sonar.server.db.migrations.MassUpdate;
 import org.sonar.server.db.migrations.Select.Row;
 import org.sonar.server.db.migrations.Select.RowReader;
+import org.sonar.server.db.migrations.SqlStatement;
 
 import java.sql.SQLException;
 import java.util.Date;
 
-
 /**
  * Used in the Active Record Migration 714
  *
@@ -56,14 +57,15 @@ public class FeedFileSources extends BaseDataChange {
       byte[] shortDates = row.getBytes(9);
       byte[] longDates = row.getBytes(10);
 
-      String sourceData = new FileSourceDto(source, shortRevisions, longRevisions, shortAuthors, longAuthors, shortDates, longDates).getSourceData();
+      String[] sourceData = new FileSourceDto(source, shortRevisions, longRevisions, shortAuthors, longAuthors, shortDates, longDates).getSourceData();
 
       update.setString(1, projectUuid)
         .setString(2, fileUuid)
         .setLong(3, now.getTime())
         .setLong(4, (updatedAt == null ? now : updatedAt).getTime())
-        .setString(5, sourceData)
-        .setString(6, "");
+        .setString(5, sourceData[0])
+        .setString(6, sourceData[1])
+        .setString(7, "");
 
       return true;
     }
@@ -91,43 +93,43 @@ public class FeedFileSources extends BaseDataChange {
 
     MassUpdate massUpdate = context.prepareMassUpdate();
     massUpdate.select("SELECT " +
-        "p.uuid as project_uuid, " +
-        "f.uuid as file_uuid, " +
-        "ss.data as source, " +
-        "ss.updated_at, " +
-        "m1.text_value as short_revisions_by_line, " +
-        "m1.measure_data as long_revisions_by_line, " +
-        "m2.text_value as short_authors_by_line, " +
-        "m2.measure_data as long_authors_by_line, " +
-        "m3.text_value as short_dates_by_line, " +
-        "m3.measure_data as short_dates_by_line " +
+      "p.uuid as project_uuid, " +
+      "f.uuid as file_uuid, " +
+      "ss.data as source, " +
+      "ss.updated_at, " +
+      "m1.text_value as short_revisions_by_line, " +
+      "m1.measure_data as long_revisions_by_line, " +
+      "m2.text_value as short_authors_by_line, " +
+      "m2.measure_data as long_authors_by_line, " +
+      "m3.text_value as short_dates_by_line, " +
+      "m3.measure_data as short_dates_by_line " +
       "FROM snapshots s " +
       "JOIN snapshot_sources ss " +
-        "ON s.id = ss.snapshot_id AND s.islast = ? " +
+      "ON s.id = ss.snapshot_id AND s.islast = ? " +
       "JOIN projects p " +
-        "ON s.root_project_id = p.id " +
+      "ON s.root_project_id = p.id " +
       "JOIN projects f " +
-        "ON s.project_id = f.id " +
+      "ON s.project_id = f.id " +
       "LEFT JOIN project_measures m1 " +
-        "ON m1.snapshot_id = s.id AND m1.metric_id = ? " +
+      "ON m1.snapshot_id = s.id AND m1.metric_id = ? " +
       "LEFT JOIN project_measures m2 " +
-        "ON m2.snapshot_id = s.id AND m2.metric_id = ? " +
+      "ON m2.snapshot_id = s.id AND m2.metric_id = ? " +
       "LEFT JOIN project_measures m3 " +
-        "ON m3.snapshot_id = s.id AND m3.metric_id = ? " +
+      "ON m3.snapshot_id = s.id AND m3.metric_id = ? " +
       "WHERE " +
-        "f.enabled = ? " +
-        "AND f.scope = 'FIL' " +
-        "AND p.scope = 'PRJ' AND p.qualifier = 'TRK' ")
-        .setBoolean(1, true)
-        .setLong(2, revisionMetricId != null ? revisionMetricId : 0L)
-        .setLong(3, authorMetricId != null ? authorMetricId : 0L)
-        .setLong(4, datesMetricId != null ? datesMetricId : 0L)
-        .setBoolean(5, true);
+      "f.enabled = ? " +
+      "AND f.scope = 'FIL' " +
+      "AND p.scope = 'PRJ' AND p.qualifier = 'TRK' ")
+      .setBoolean(1, true)
+      .setLong(2, revisionMetricId != null ? revisionMetricId : 0L)
+      .setLong(3, authorMetricId != null ? authorMetricId : 0L)
+      .setLong(4, datesMetricId != null ? datesMetricId : 0L)
+      .setBoolean(5, true);
 
     massUpdate.update("INSERT INTO file_sources" +
-      "(project_uuid, file_uuid, created_at, updated_at, data, data_hash)" +
+      "(project_uuid, file_uuid, created_at, updated_at, data, line_hashes, data_hash)" +
       "VALUES " +
-      "(?, ?, ?, ?, ?, ?)");
+      "(?, ?, ?, ?, ?, ?, ?)");
 
     massUpdate.execute(new FileSourceBuilder(system));
   }
diff --git a/server/sonar-server/src/main/java/org/sonar/server/db/migrations/v50/FileSourceDto.java b/server/sonar-server/src/main/java/org/sonar/server/db/migrations/v50/FileSourceDto.java
index 893e1fc8c6b..31a5ec2849f 100644
--- a/server/sonar-server/src/main/java/org/sonar/server/db/migrations/v50/FileSourceDto.java
+++ b/server/sonar-server/src/main/java/org/sonar/server/db/migrations/v50/FileSourceDto.java
@@ -20,6 +20,8 @@
 package org.sonar.server.db.migrations.v50;
 
 import com.google.common.base.Splitter;
+import org.apache.commons.codec.digest.DigestUtils;
+import org.apache.commons.lang.StringUtils;
 import org.sonar.api.utils.KeyValueFormat;
 import org.sonar.api.utils.text.CsvWriter;
 
@@ -34,6 +36,8 @@ import static com.google.common.base.Charsets.UTF_8;
 
 class FileSourceDto {
 
+  private static final String SPACE_CHARS = "\t\n\r ";
+
   private Iterator<String> sourceSplitter;
 
   private Map<Integer, String> revisions;
@@ -50,19 +54,29 @@ class FileSourceDto {
     dates = KeyValueFormat.parseIntString(ofNullableBytes(shortDates, longDates));
   }
 
-  String getSourceData() {
+  String[] getSourceData() {
     String highlighting = "";
     ByteArrayOutputStream output = new ByteArrayOutputStream();
     int line = 0;
     String sourceLine = null;
     CsvWriter csv = CsvWriter.of(new OutputStreamWriter(output, UTF_8));
+    StringBuilder lineHashes = new StringBuilder();
     while (sourceSplitter.hasNext()) {
-      line ++;
+      line++;
       sourceLine = sourceSplitter.next();
+      lineHashes.append(lineChecksum(sourceLine)).append("\n");
       csv.values(revisions.get(line), authors.get(line), dates.get(line), highlighting, sourceLine);
     }
     csv.close();
-    return new String(output.toByteArray(), UTF_8);
+    return new String[] {new String(output.toByteArray(), UTF_8), lineHashes.toString()};
+  }
+
+  public static String lineChecksum(String line) {
+    String reducedLine = StringUtils.replaceChars(line, SPACE_CHARS, "");
+    if (line.isEmpty()) {
+      return "";
+    }
+    return DigestUtils.md5Hex(reducedLine);
   }
 
   private static String ofNullableBytes(@Nullable byte[] shortBytes, @Nullable byte[] longBytes) {
diff --git a/server/sonar-web/src/main/webapp/WEB-INF/db/migrate/713_create_file_sources.rb b/server/sonar-web/src/main/webapp/WEB-INF/db/migrate/713_create_file_sources.rb
index 422f355f95b..65775fc9adf 100644
--- a/server/sonar-web/src/main/webapp/WEB-INF/db/migrate/713_create_file_sources.rb
+++ b/server/sonar-web/src/main/webapp/WEB-INF/db/migrate/713_create_file_sources.rb
@@ -28,6 +28,7 @@ class CreateFileSources < ActiveRecord::Migration
       t.column :project_uuid, :string,   :limit => 50, :null => false
       t.column :file_uuid,    :string,   :limit => 50, :null => false
       t.column :data,         :text,     :null => true
+      t.column :line_hashes,  :text,     :null => true
       t.column :data_hash,    :string,   :limit => 50, :null => true
       t.column :created_at,   :integer,  :limit => 8, :null => false 
       t.column :updated_at,   :integer,  :limit => 8, :null => false
diff --git a/sonar-batch/src/main/java/org/sonar/batch/index/SourcePersister.java b/sonar-batch/src/main/java/org/sonar/batch/index/SourcePersister.java
index 6e42a07c18e..d2d27beb498 100644
--- a/sonar-batch/src/main/java/org/sonar/batch/index/SourcePersister.java
+++ b/sonar-batch/src/main/java/org/sonar/batch/index/SourcePersister.java
@@ -20,6 +20,7 @@
 package org.sonar.batch.index;
 
 import com.google.common.base.CharMatcher;
+import com.google.common.base.Joiner;
 import org.apache.commons.codec.digest.DigestUtils;
 import org.apache.commons.io.FileUtils;
 import org.apache.commons.lang.StringUtils;
@@ -56,7 +57,12 @@ import javax.annotation.Nullable;
 import java.io.ByteArrayOutputStream;
 import java.io.IOException;
 import java.io.OutputStreamWriter;
-import java.util.*;
+import java.util.Collections;
+import java.util.Date;
+import java.util.HashMap;
+import java.util.Iterator;
+import java.util.List;
+import java.util.Map;
 
 import static com.google.common.base.Charsets.UTF_8;
 
@@ -147,15 +153,23 @@ public class SourcePersister implements ScanPersister {
     String newDataHash = newData != null ? DigestUtils.md5Hex(newData) : "0";
     Date now = system2.newDate();
     if (previous == null) {
-      FileSourceDto newFileSource = new FileSourceDto().setProjectUuid(projectTree.getRootProject().getUuid()).setFileUuid(fileUuid).setData(newData)
+      FileSourceDto newFileSource = new FileSourceDto()
+        .setProjectUuid(projectTree.getRootProject().getUuid())
+        .setFileUuid(fileUuid)
+        .setData(newData)
         .setDataHash(newDataHash)
+        .setLineHashes(StringUtils.defaultIfEmpty(Joiner.on('\n').join(inputFile.lineHashes()), null))
         .setCreatedAt(now.getTime())
         .setUpdatedAt(now.getTime());
       mapper.insert(newFileSource);
       session.commit();
     } else {
       if (!newDataHash.equals(previous.getDataHash())) {
-        previous.setData(newData).setDataHash(newDataHash).setUpdatedAt(now.getTime());
+        previous
+          .setData(newData)
+          .setLineHashes(StringUtils.defaultIfEmpty(Joiner.on('\n').join(inputFile.lineHashes()), null))
+          .setDataHash(newDataHash)
+          .setUpdatedAt(now.getTime());
         mapper.update(previous);
         session.commit();
       }
diff --git a/sonar-batch/src/main/java/org/sonar/batch/scan/filesystem/DefaultInputFileValueCoder.java b/sonar-batch/src/main/java/org/sonar/batch/scan/filesystem/DefaultInputFileValueCoder.java
index 49b19f6fec1..e216534167f 100644
--- a/sonar-batch/src/main/java/org/sonar/batch/scan/filesystem/DefaultInputFileValueCoder.java
+++ b/sonar-batch/src/main/java/org/sonar/batch/scan/filesystem/DefaultInputFileValueCoder.java
@@ -48,6 +48,7 @@ class DefaultInputFileValueCoder implements ValueCoder {
     value.put(f.lines());
     putUTFOrNull(value, f.encoding());
     value.putLongArray(f.originalLineOffsets());
+    value.putStringArray(f.lineHashes());
   }
 
   private void putUTFOrNull(Value value, @Nullable String utfOrNull) {
@@ -74,6 +75,7 @@ class DefaultInputFileValueCoder implements ValueCoder {
     file.setLines(value.getInt());
     file.setEncoding(value.getString());
     file.setOriginalLineOffsets(value.getLongArray());
+    file.setLineHashes(value.getStringArray());
     return file;
   }
 
diff --git a/sonar-batch/src/main/java/org/sonar/batch/scan/filesystem/FileMetadata.java b/sonar-batch/src/main/java/org/sonar/batch/scan/filesystem/FileMetadata.java
index ca6fd9aed4b..95089291ef6 100644
--- a/sonar-batch/src/main/java/org/sonar/batch/scan/filesystem/FileMetadata.java
+++ b/sonar-batch/src/main/java/org/sonar/batch/scan/filesystem/FileMetadata.java
@@ -19,11 +19,11 @@
  */
 package org.sonar.batch.scan.filesystem;
 
-import com.google.common.primitives.Ints;
 import com.google.common.primitives.Longs;
 import org.apache.commons.codec.binary.Hex;
 import org.apache.commons.codec.digest.DigestUtils;
 import org.apache.commons.io.IOUtils;
+import org.apache.commons.lang.StringUtils;
 
 import java.io.BufferedReader;
 import java.io.File;
@@ -45,6 +45,7 @@ class FileMetadata {
   private static final char LINE_FEED = '\n';
   private static final char CARRIAGE_RETURN = '\r';
   private static final char BOM = '\uFEFF';
+  private static final String SPACE_CHARS = "\t\n\r ";
 
   // This singleton aims only to increase the coverage by allowing
   // to test the private method !
@@ -61,14 +62,13 @@ class FileMetadata {
     Reader reader = null;
     long currentOriginalOffset = 0;
     List<Long> originalLineOffsets = new ArrayList<Long>();
-    List<Integer> lineCheckSum = new ArrayList<Integer>();
-    int hash = 5381;
+    List<String> lineHashes = new ArrayList<String>();
     StringBuilder currentLineStr = new StringBuilder();
     int lines = 0;
     char c = (char) -1;
     try {
-      MessageDigest md5Digest = DigestUtils.getMd5Digest();
-      md5Digest.reset();
+      MessageDigest globalMd5Digest = DigestUtils.getMd5Digest();
+      globalMd5Digest.reset();
       reader = new BufferedReader(new InputStreamReader(new FileInputStream(file), encoding));
       int i = reader.read();
       boolean afterCR = false;
@@ -95,24 +95,24 @@ class FileMetadata {
           afterCR = true;
           c = LINE_FEED;
         }
-        currentLineStr.append(c);
-        hash = ((hash << 5) + hash) + (c & 0xff);
         if (c == LINE_FEED) {
           lines++;
           originalLineOffsets.add(currentOriginalOffset);
-          lineCheckSum.add(hash);
-          hash = 5381;
+          lineHashes.add(md5IgnoreWhitespace(currentLineStr));
           currentLineStr.setLength(0);
+        } else {
+          currentLineStr.append(c);
         }
-        md5Digest.update(charToBytesUTF(c));
+        globalMd5Digest.update(charToBytesUTF(c));
         i = reader.read();
       }
       if (c != (char) -1) {
+        // Last empty line
         lines++;
-        lineCheckSum.add(hash);
+        lineHashes.add(md5IgnoreWhitespace(currentLineStr));
       }
-      String filehash = Hex.encodeHexString(md5Digest.digest());
-      return new Metadata(lines, filehash, originalLineOffsets, lineCheckSum);
+      String filehash = Hex.encodeHexString(globalMd5Digest.digest());
+      return new Metadata(lines, filehash, originalLineOffsets, lineHashes.toArray(new String[0]));
 
     } catch (IOException e) {
       throw new IllegalStateException(String.format("Fail to read file '%s' with encoding '%s'", file.getAbsolutePath(), encoding), e);
@@ -121,6 +121,14 @@ class FileMetadata {
     }
   }
 
+  private String md5IgnoreWhitespace(StringBuilder currentLineStr) {
+    String reducedLine = StringUtils.replaceChars(currentLineStr.toString(), SPACE_CHARS, "");
+    if (reducedLine.isEmpty()) {
+      return "";
+    }
+    return DigestUtils.md5Hex(reducedLine);
+  }
+
   private byte[] charToBytesUTF(char c) {
     char[] buffer = new char[] {c};
     byte[] b = new byte[buffer.length << 1];
@@ -136,13 +144,13 @@ class FileMetadata {
     final int lines;
     final String hash;
     final long[] originalLineOffsets;
-    final int[] lineChecksum;
+    final String[] lineHashes;
 
-    private Metadata(int lines, String hash, List<Long> originalLineOffsets, List<Integer> lineCheckSum) {
+    private Metadata(int lines, String hash, List<Long> originalLineOffsets, String[] lineHashes) {
       this.lines = lines;
       this.hash = hash;
       this.originalLineOffsets = Longs.toArray(originalLineOffsets);
-      this.lineChecksum = Ints.toArray(lineCheckSum);
+      this.lineHashes = lineHashes;
     }
   }
 }
diff --git a/sonar-batch/src/main/java/org/sonar/batch/scan/filesystem/InputFileBuilder.java b/sonar-batch/src/main/java/org/sonar/batch/scan/filesystem/InputFileBuilder.java
index a94f059e85a..b961e072890 100644
--- a/sonar-batch/src/main/java/org/sonar/batch/scan/filesystem/InputFileBuilder.java
+++ b/sonar-batch/src/main/java/org/sonar/batch/scan/filesystem/InputFileBuilder.java
@@ -100,6 +100,7 @@ class InputFileBuilder {
     inputFile.setLines(metadata.lines);
     inputFile.setHash(metadata.hash);
     inputFile.setOriginalLineOffsets(metadata.originalLineOffsets);
+    inputFile.setLineHashes(metadata.lineHashes);
     inputFile.setStatus(statusDetection.status(inputFile.moduleKey(), inputFile.relativePath(), metadata.hash));
     if (analysisMode.isIncremental() && inputFile.status() == InputFile.Status.SAME) {
       return null;
diff --git a/sonar-batch/src/test/java/org/sonar/batch/index/SourcePersisterTest.java b/sonar-batch/src/test/java/org/sonar/batch/index/SourcePersisterTest.java
index 8cb3ec22724..7452dc2995e 100644
--- a/sonar-batch/src/test/java/org/sonar/batch/index/SourcePersisterTest.java
+++ b/sonar-batch/src/test/java/org/sonar/batch/index/SourcePersisterTest.java
@@ -130,7 +130,9 @@ public class SourcePersisterTest extends AbstractDaoTestCase {
     String relativePathSame = "src/changed.java";
     java.io.File sameFile = new java.io.File(basedir, relativePathSame);
     FileUtils.write(sameFile, "changed\ncontent");
-    DefaultInputFile inputFileNew = new DefaultInputFile(PROJECT_KEY, relativePathSame).setLines(2).setAbsolutePath(sameFile.getAbsolutePath());
+    DefaultInputFile inputFileNew = new DefaultInputFile(PROJECT_KEY, relativePathSame).setLines(2)
+      .setAbsolutePath(sameFile.getAbsolutePath())
+      .setLineHashes(new String[] {"foo", "bar"});
     when(inputPathCache.all()).thenReturn(Arrays.<InputPath>asList(inputFileNew));
 
     mockResourceCache(relativePathSame, PROJECT_KEY, "uuidsame");
@@ -142,6 +144,7 @@ public class SourcePersisterTest extends AbstractDaoTestCase {
     assertThat(fileSourceDto.getUpdatedAt()).isEqualTo(now.getTime());
     assertThat(fileSourceDto.getData()).isEqualTo(
       ",,,,changed\r\n,,,,content\r\n");
+    assertThat(fileSourceDto.getLineHashes()).isEqualTo("foo\nbar");
     assertThat(fileSourceDto.getDataHash()).isEqualTo("e41cca9c51ff853c748f708f39dfc035");
   }
 
@@ -151,7 +154,9 @@ public class SourcePersisterTest extends AbstractDaoTestCase {
     when(system2.newDate()).thenReturn(DateUtils.parseDateTime("2014-10-29T16:44:02+0100"));
 
     String relativePathEmpty = "src/empty.java";
-    DefaultInputFile inputFileEmpty = new DefaultInputFile(PROJECT_KEY, relativePathEmpty).setLines(0);
+    DefaultInputFile inputFileEmpty = new DefaultInputFile(PROJECT_KEY, relativePathEmpty)
+      .setLines(0)
+      .setLineHashes(new String[] {});
     when(inputPathCache.all()).thenReturn(Arrays.<InputPath>asList(inputFileEmpty));
 
     mockResourceCache(relativePathEmpty, PROJECT_KEY, "uuidempty");
@@ -169,7 +174,10 @@ public class SourcePersisterTest extends AbstractDaoTestCase {
     String relativePathNew = "src/new.java";
     java.io.File newFile = new java.io.File(basedir, relativePathNew);
     FileUtils.write(newFile, "foo\nbar\nbiz");
-    DefaultInputFile inputFileNew = new DefaultInputFile(PROJECT_KEY, relativePathNew).setLines(3).setAbsolutePath(newFile.getAbsolutePath());
+    DefaultInputFile inputFileNew = new DefaultInputFile(PROJECT_KEY, relativePathNew)
+      .setLines(3)
+      .setAbsolutePath(newFile.getAbsolutePath())
+      .setLineHashes(new String[] {"foo", "bar", "bee"});
     when(inputPathCache.all()).thenReturn(Arrays.<InputPath>asList(inputFileNew));
 
     mockResourceCache(relativePathNew, PROJECT_KEY, "uuidnew");
@@ -180,6 +188,7 @@ public class SourcePersisterTest extends AbstractDaoTestCase {
     assertThat(fileSourceDto.getUpdatedAt()).isEqualTo(now.getTime());
     assertThat(fileSourceDto.getData()).isEqualTo(
       ",,,,foo\r\n,,,,bar\r\n,,,,biz\r\n");
+    assertThat(fileSourceDto.getLineHashes()).isEqualTo("foo\nbar\nbee");
     assertThat(fileSourceDto.getDataHash()).isEqualTo("0c43ed6418d690ee0ffc3e43e6660967");
 
   }
@@ -196,7 +205,8 @@ public class SourcePersisterTest extends AbstractDaoTestCase {
     DefaultInputFile inputFileNew = new DefaultInputFile(PROJECT_KEY, relativePathNew)
       .setLines(3)
       .setAbsolutePath(newFile.getAbsolutePath())
-      .setOriginalLineOffsets(new long[] {0, 4, 7});
+      .setOriginalLineOffsets(new long[] {0, 4, 7})
+      .setLineHashes(new String[] {"foo", "bar", "bee"});
     when(inputPathCache.all()).thenReturn(Arrays.<InputPath>asList(inputFileNew));
 
     mockResourceCache(relativePathNew, PROJECT_KEY, "uuidnew");
@@ -221,6 +231,7 @@ public class SourcePersisterTest extends AbstractDaoTestCase {
     FileSourceDto fileSourceDto = new FileSourceDao(getMyBatis()).select("uuidnew");
     assertThat(fileSourceDto.getCreatedAt()).isEqualTo(now.getTime());
     assertThat(fileSourceDto.getUpdatedAt()).isEqualTo(now.getTime());
+    assertThat(fileSourceDto.getLineHashes()).isEqualTo("foo\nbar\nbee");
     assertThat(fileSourceDto.getData()).isEqualTo(
       "123,julien,2014-10-11T16:44:02+0100,\"0,3,a\",foo\r\n"
         + "234,simon,2014-10-12T16:44:02+0100,\"0,1,cd\",bar\r\n"
diff --git a/sonar-batch/src/test/java/org/sonar/batch/scan/filesystem/FileMetadataTest.java b/sonar-batch/src/test/java/org/sonar/batch/scan/filesystem/FileMetadataTest.java
index f3a1d51c990..6814f6888ed 100644
--- a/sonar-batch/src/test/java/org/sonar/batch/scan/filesystem/FileMetadataTest.java
+++ b/sonar-batch/src/test/java/org/sonar/batch/scan/filesystem/FileMetadataTest.java
@@ -20,6 +20,7 @@
 package org.sonar.batch.scan.filesystem;
 
 import com.google.common.base.Charsets;
+import org.apache.commons.codec.digest.DigestUtils;
 import org.apache.commons.io.FileUtils;
 import org.junit.Rule;
 import org.junit.Test;
@@ -52,7 +53,7 @@ public class FileMetadataTest {
     assertThat(metadata.lines).isEqualTo(0);
     assertThat(metadata.hash).isNotEmpty();
     assertThat(metadata.originalLineOffsets).containsOnly(0);
-    assertThat(metadata.lineChecksum).isEmpty();
+    assertThat(metadata.lineHashes).isEmpty();
   }
 
   @Test
@@ -64,7 +65,7 @@ public class FileMetadataTest {
     assertThat(metadata.lines).isEqualTo(3);
     assertThat(metadata.hash).isEqualTo(EXPECTED_HASH_WITHOUT_LATEST_EOL);
     assertThat(metadata.originalLineOffsets).containsOnly(0, 5, 10);
-    assertThat(metadata.lineChecksum).containsOnly(2090263731, 2090104836, 193487042);
+    assertThat(metadata.lineHashes).containsOnly(md5("foo"), md5("bar"), md5("baz"));
   }
 
   @Test
@@ -76,7 +77,7 @@ public class FileMetadataTest {
     assertThat(metadata.lines).isEqualTo(4);
     assertThat(metadata.hash).isEqualTo(NON_ASCII);
     assertThat(metadata.originalLineOffsets).containsOnly(0, 5, 10, 18);
-    assertThat(metadata.lineChecksum).containsOnly(2090410746, 2090243139, -931663839, 5381);
+    assertThat(metadata.lineHashes).containsOnly(md5("föo"), md5("bàr"), md5("\u1D11Ebaßz"), "");
   }
 
   @Test
@@ -88,7 +89,7 @@ public class FileMetadataTest {
     assertThat(metadata.lines).isEqualTo(4);
     assertThat(metadata.hash).isEqualTo(NON_ASCII);
     assertThat(metadata.originalLineOffsets).containsOnly(0, 5, 10, 18);
-    assertThat(metadata.lineChecksum).containsOnly(2090410746, 2090243139, -931663839, 5381);
+    assertThat(metadata.lineHashes).containsOnly(md5("föo"), md5("bàr"), md5("\u1D11Ebaßz"), "");
   }
 
   @Test
@@ -100,7 +101,7 @@ public class FileMetadataTest {
     assertThat(metadata.lines).isEqualTo(3);
     assertThat(metadata.hash).isEqualTo(EXPECTED_HASH_WITHOUT_LATEST_EOL);
     assertThat(metadata.originalLineOffsets).containsOnly(0, 4, 8);
-    assertThat(metadata.lineChecksum).containsOnly(2090263731, 2090104836, 193487042);
+    assertThat(metadata.lineHashes).containsOnly(md5("foo"), md5("bar"), md5("baz"));
   }
 
   @Test
@@ -112,7 +113,7 @@ public class FileMetadataTest {
     assertThat(metadata.lines).isEqualTo(4);
     assertThat(metadata.hash).isEqualTo(EXPECTED_HASH_WITH_LATEST_EOL);
     assertThat(metadata.originalLineOffsets).containsOnly(0, 4, 8, 12);
-    assertThat(metadata.lineChecksum).containsOnly(2090263731, 2090104836, 2090105100, 5381);
+    assertThat(metadata.lineHashes).containsOnly(md5("foo"), md5("bar"), md5("baz"), "");
   }
 
   @Test
@@ -124,7 +125,7 @@ public class FileMetadataTest {
     assertThat(metadata.lines).isEqualTo(4);
     assertThat(metadata.hash).isEqualTo(EXPECTED_HASH_WITH_LATEST_EOL);
     assertThat(metadata.originalLineOffsets).containsOnly(0, 4, 9, 13);
-    assertThat(metadata.lineChecksum).containsOnly(2090263731, 2090104836, 2090105100, 5381);
+    assertThat(metadata.lineHashes).containsOnly(md5("foo"), md5("bar"), md5("baz"), "");
   }
 
   @Test
@@ -136,7 +137,7 @@ public class FileMetadataTest {
     assertThat(metadata.lines).isEqualTo(3);
     assertThat(metadata.hash).isEqualTo(EXPECTED_HASH_WITHOUT_LATEST_EOL);
     assertThat(metadata.originalLineOffsets).containsOnly(0, 4, 9);
-    assertThat(metadata.lineChecksum).containsOnly(2090263731, 2090104836, 193487042);
+    assertThat(metadata.lineHashes).containsOnly(md5("foo"), md5("bar"), md5("baz"));
   }
 
   @Test
@@ -148,7 +149,7 @@ public class FileMetadataTest {
     assertThat(metadata.lines).isEqualTo(4);
     assertThat(metadata.hash).isEqualTo(EXPECTED_HASH_NEW_LINE_FIRST);
     assertThat(metadata.originalLineOffsets).containsOnly(0, 1, 5, 10);
-    assertThat(metadata.lineChecksum).containsOnly(177583, 2090263731, 2090104836, 193487042);
+    assertThat(metadata.lineHashes).containsOnly("", md5("foo"), md5("bar"), md5("baz"));
   }
 
   @Test
@@ -160,7 +161,17 @@ public class FileMetadataTest {
     assertThat(metadata.lines).isEqualTo(3);
     assertThat(metadata.hash).isEqualTo(EXPECTED_HASH_WITHOUT_LATEST_EOL);
     assertThat(metadata.originalLineOffsets).containsOnly(0, 4, 9);
-    assertThat(metadata.lineChecksum).containsOnly(2090263731, 2090104836, 193487042);
+    assertThat(metadata.lineHashes).containsOnly(md5("foo"), md5("bar"), md5("baz"));
+  }
+
+  @Test
+  public void ignore_whitespace_when_computing_line_hashes() throws Exception {
+    File tempFile = temp.newFile();
+    FileUtils.write(tempFile, " foo\nb ar\r\nbaz \t", Charsets.UTF_8, true);
+
+    FileMetadata.Metadata metadata = FileMetadata.INSTANCE.read(tempFile, Charsets.UTF_8);
+    assertThat(metadata.lines).isEqualTo(3);
+    assertThat(metadata.lineHashes).containsOnly(md5("foo"), md5("bar"), md5("baz"));
   }
 
   @Test
@@ -192,4 +203,8 @@ public class FileMetadataTest {
     assertThat(hash1).isEqualTo(hash1a);
     assertThat(hash1).isNotEqualTo(hash2);
   }
+
+  private static String md5(String input) {
+    return DigestUtils.md5Hex(input);
+  }
 }
diff --git a/sonar-batch/src/test/java/org/sonar/batch/scan/filesystem/InputPathCacheTest.java b/sonar-batch/src/test/java/org/sonar/batch/scan/filesystem/InputPathCacheTest.java
index 7008b0450c0..bbb63290841 100644
--- a/sonar-batch/src/test/java/org/sonar/batch/scan/filesystem/InputPathCacheTest.java
+++ b/sonar-batch/src/test/java/org/sonar/batch/scan/filesystem/InputPathCacheTest.java
@@ -67,10 +67,16 @@ public class InputPathCacheTest {
       .setStatus(Status.ADDED)
       .setHash("xyz")
       .setLines(1)
+      .setEncoding("UTF-8")
+      .setOriginalLineOffsets(new long[] {0, 4})
+      .setLineHashes(new String[] {"foo", "bar"})
       .setFile(temp.newFile("Bar.java")));
 
-    assertThat(cache.getFile("struts", "src/main/java/Foo.java").relativePath())
-      .isEqualTo("src/main/java/Foo.java");
+    DefaultInputFile loadedFile = (DefaultInputFile) cache.getFile("struts-core", "src/main/java/Bar.java");
+    assertThat(loadedFile.relativePath()).isEqualTo("src/main/java/Bar.java");
+    assertThat(loadedFile.encoding()).isEqualTo("UTF-8");
+    assertThat(loadedFile.originalLineOffsets()).containsOnly(0, 4);
+    assertThat(loadedFile.lineHashes()).containsOnly("foo", "bar");
 
     assertThat(cache.filesByModule("struts")).hasSize(1);
     assertThat(cache.filesByModule("struts-core")).hasSize(1);
diff --git a/sonar-batch/src/test/resources/org/sonar/batch/index/SourcePersisterTest/file_sources.xml b/sonar-batch/src/test/resources/org/sonar/batch/index/SourcePersisterTest/file_sources.xml
index fae5b462879..a4e8ed467b3 100644
--- a/sonar-batch/src/test/resources/org/sonar/batch/index/SourcePersisterTest/file_sources.xml
+++ b/sonar-batch/src/test/resources/org/sonar/batch/index/SourcePersisterTest/file_sources.xml
@@ -1,5 +1,8 @@
 <dataset>
-
-  <file_sources id="101" project_uuid="projectUuid" file_uuid="uuidsame" data=",,,,unchanged&#13;&#10;,,,,content&#13;&#10;" data_hash="ee716d4ed9faae16eb9167714442a3bc" created_at="1412952242000" updated_at="1412952242000" />
+  <file_sources id="101" project_uuid="projectUuid" file_uuid="uuidsame" 
+      data=",,,,unchanged&#13;&#10;,,,,content&#13;&#10;" 
+      line_hashes="8d7b3d6b83c0a517eac07e1aac94b773&#10;9a0364b9e99bb480dd25e1f0284c8555" 
+      data_hash="ee716d4ed9faae16eb9167714442a3bc" 
+      created_at="1412952242000" updated_at="1412952242000" />
 
 </dataset>
diff --git a/sonar-batch/src/test/resources/org/sonar/batch/index/SourcePersisterTest/testPersistDontTouchUnchanged-result.xml b/sonar-batch/src/test/resources/org/sonar/batch/index/SourcePersisterTest/testPersistDontTouchUnchanged-result.xml
index fae5b462879..98dd81f73df 100644
--- a/sonar-batch/src/test/resources/org/sonar/batch/index/SourcePersisterTest/testPersistDontTouchUnchanged-result.xml
+++ b/sonar-batch/src/test/resources/org/sonar/batch/index/SourcePersisterTest/testPersistDontTouchUnchanged-result.xml
@@ -1,5 +1,9 @@
 <dataset>
 
-  <file_sources id="101" project_uuid="projectUuid" file_uuid="uuidsame" data=",,,,unchanged&#13;&#10;,,,,content&#13;&#10;" data_hash="ee716d4ed9faae16eb9167714442a3bc" created_at="1412952242000" updated_at="1412952242000" />
+  <file_sources id="101" project_uuid="projectUuid" file_uuid="uuidsame" 
+      data=",,,,unchanged&#13;&#10;,,,,content&#13;&#10;" 
+      line_hashes="8d7b3d6b83c0a517eac07e1aac94b773&#10;9a0364b9e99bb480dd25e1f0284c8555" 
+      data_hash="ee716d4ed9faae16eb9167714442a3bc" 
+      created_at="1412952242000" updated_at="1412952242000" />
 
 </dataset>
diff --git a/sonar-batch/src/test/resources/org/sonar/batch/index/SourcePersisterTest/testPersistEmptyFile-result.xml b/sonar-batch/src/test/resources/org/sonar/batch/index/SourcePersisterTest/testPersistEmptyFile-result.xml
index ac3c16563af..cd844a149d8 100644
--- a/sonar-batch/src/test/resources/org/sonar/batch/index/SourcePersisterTest/testPersistEmptyFile-result.xml
+++ b/sonar-batch/src/test/resources/org/sonar/batch/index/SourcePersisterTest/testPersistEmptyFile-result.xml
@@ -1,6 +1,13 @@
 <dataset>
-    <file_sources id="101" project_uuid="projectUuid" file_uuid="uuidsame" data=",,,,unchanged&#13;&#10;,,,,content&#13;&#10;" data_hash="ee716d4ed9faae16eb9167714442a3bc" created_at="1412952242000" updated_at="1412952242000" />
-    <file_sources id="102" project_uuid="projectUuid" file_uuid="uuidempty" data="[null]" data_hash="0" created_at="1414597442000" updated_at="1414597442000" />
+    <file_sources id="101" project_uuid="projectUuid" file_uuid="uuidsame" 
+      data=",,,,unchanged&#13;&#10;,,,,content&#13;&#10;" 
+      line_hashes="8d7b3d6b83c0a517eac07e1aac94b773&#10;9a0364b9e99bb480dd25e1f0284c8555" 
+      data_hash="ee716d4ed9faae16eb9167714442a3bc" 
+      created_at="1412952242000" updated_at="1412952242000" />
+      
+    <file_sources id="102" project_uuid="projectUuid" file_uuid="uuidempty" data="[null]"
+       line_hashes="[null]" 
+       data_hash="0" created_at="1414597442000" updated_at="1414597442000" />
 </dataset>
 
 
diff --git a/sonar-core/src/main/java/org/sonar/core/source/db/FileSourceDto.java b/sonar-core/src/main/java/org/sonar/core/source/db/FileSourceDto.java
index e992ee4a648..722f10551fb 100644
--- a/sonar-core/src/main/java/org/sonar/core/source/db/FileSourceDto.java
+++ b/sonar-core/src/main/java/org/sonar/core/source/db/FileSourceDto.java
@@ -29,6 +29,7 @@ public class FileSourceDto {
   private long createdAt;
   private long updatedAt;
   private String data;
+  private String lineHashes;
   private String dataHash;
 
   public Long getId() {
@@ -68,6 +69,16 @@ public class FileSourceDto {
     return this;
   }
 
+  @CheckForNull
+  public String getLineHashes() {
+    return lineHashes;
+  }
+
+  public FileSourceDto setLineHashes(@Nullable String lineHashes) {
+    this.lineHashes = lineHashes;
+    return this;
+  }
+
   public String getDataHash() {
     return dataHash;
   }
diff --git a/sonar-core/src/main/resources/org/sonar/core/persistence/schema-h2.ddl b/sonar-core/src/main/resources/org/sonar/core/persistence/schema-h2.ddl
index 7de7f8d46c2..fec355edc63 100644
--- a/sonar-core/src/main/resources/org/sonar/core/persistence/schema-h2.ddl
+++ b/sonar-core/src/main/resources/org/sonar/core/persistence/schema-h2.ddl
@@ -571,6 +571,7 @@ CREATE TABLE "FILE_SOURCES" (
   "PROJECT_UUID" VARCHAR(50) NOT NULL,
   "FILE_UUID" VARCHAR(50) NOT NULL,
   "DATA" CLOB(2147483647),
+  "LINE_HASHES" CLOB(2147483647),
   "DATA_HASH" VARCHAR(50) NOT NULL,
   "CREATED_AT" BIGINT NOT NULL,
   "UPDATED_AT" BIGINT NOT NULL
diff --git a/sonar-core/src/main/resources/org/sonar/core/source/db/FileSourceMapper.xml b/sonar-core/src/main/resources/org/sonar/core/source/db/FileSourceMapper.xml
index 4b3b88465c3..d7b65c501ea 100644
--- a/sonar-core/src/main/resources/org/sonar/core/source/db/FileSourceMapper.xml
+++ b/sonar-core/src/main/resources/org/sonar/core/source/db/FileSourceMapper.xml
@@ -5,7 +5,7 @@
 <mapper namespace="org.sonar.core.source.db.FileSourceMapper">
 
   <select id="select" parameterType="string" resultType="org.sonar.core.source.db.FileSourceDto">
-    SELECT id, project_uuid as projectUuid, file_uuid as fileUuid, created_at as createdAt, updated_at as updatedAt, data, data_hash as dataHash
+    SELECT id, project_uuid as projectUuid, file_uuid as fileUuid, created_at as createdAt, updated_at as updatedAt, data, line_hashes as lineHashes, data_hash as dataHash
     FROM file_sources
     WHERE file_uuid = #{fileUuid}
   </select>
@@ -17,14 +17,15 @@
   </select>
   
   <insert id="insert" parameterType="org.sonar.core.source.db.FileSourceDto" useGeneratedKeys="false">
-    insert into file_sources (project_uuid, file_uuid, created_at, updated_at, data, data_hash) 
-    values (#{projectUuid}, #{fileUuid}, #{createdAt}, #{updatedAt}, #{data}, #{dataHash})
+    insert into file_sources (project_uuid, file_uuid, created_at, updated_at, data, line_hashes, data_hash) 
+    values (#{projectUuid}, #{fileUuid}, #{createdAt}, #{updatedAt}, #{data}, #{lineHashes}, #{dataHash})
   </insert>
   
   <update id="update" parameterType="org.sonar.core.source.db.FileSourceDto" useGeneratedKeys="false">
     update file_sources set
       updated_at = #{updatedAt},
       data = #{data},
+      line_hashes = #{lineHashes},
       data_hash = #{dataHash}
     where id = #{id}
   </update>
diff --git a/sonar-core/src/test/java/org/sonar/core/source/db/FileSourceDaoTest.java b/sonar-core/src/test/java/org/sonar/core/source/db/FileSourceDaoTest.java
index ddd45cedfa5..0487d80b379 100644
--- a/sonar-core/src/test/java/org/sonar/core/source/db/FileSourceDaoTest.java
+++ b/sonar-core/src/test/java/org/sonar/core/source/db/FileSourceDaoTest.java
@@ -53,7 +53,9 @@ public class FileSourceDaoTest extends AbstractDaoTestCase {
 
   @Test
   public void insert() throws Exception {
-    dao.insert(new FileSourceDto().setProjectUuid("prj").setFileUuid("file").setData("bla bla").setDataHash("hash2")
+    dao.insert(new FileSourceDto().setProjectUuid("prj").setFileUuid("file").setData("bla bla")
+      .setDataHash("hash2")
+      .setLineHashes("foo\nbar")
       .setCreatedAt(DateUtils.parseDateTime("2014-10-31T16:44:02+0100").getTime())
       .setUpdatedAt(DateUtils.parseDateTime("2014-10-31T16:44:02+0100").getTime()));
 
@@ -62,10 +64,12 @@ public class FileSourceDaoTest extends AbstractDaoTestCase {
 
   @Test
   public void update() throws Exception {
-    dao.update(new FileSourceDto().setId(101L).setProjectUuid("prj").setFileUuid("file").setData("updated data").setDataHash("hash2")
+    dao.update(new FileSourceDto().setId(101L).setProjectUuid("prj").setFileUuid("file")
+      .setData("updated data")
+      .setDataHash("hash2")
+      .setLineHashes("foo2\nbar2")
       .setUpdatedAt(DateUtils.parseDateTime("2014-10-31T16:44:02+0100").getTime()));
 
     checkTable("update", "file_sources");
   }
-
 }
diff --git a/sonar-core/src/test/resources/org/sonar/core/source/db/FileSourceDaoTest/insert-result.xml b/sonar-core/src/test/resources/org/sonar/core/source/db/FileSourceDaoTest/insert-result.xml
index c5d483b7e55..49ea06706e5 100644
--- a/sonar-core/src/test/resources/org/sonar/core/source/db/FileSourceDaoTest/insert-result.xml
+++ b/sonar-core/src/test/resources/org/sonar/core/source/db/FileSourceDaoTest/insert-result.xml
@@ -2,11 +2,13 @@
 
     <file_sources id="101" project_uuid="abcd" file_uuid="ab12"
                   data="aef12a,alice,2014-04-25T12:34:56+0100,,class Foo" data_hash="hash"
+                  line_hashes="truc"
                   created_at="1414597442000" updated_at="1414683842000" />
 
 
     <file_sources id="102" project_uuid="prj" file_uuid="file"
                   data="bla bla" data_hash="hash2"
+                  line_hashes="foo&#10;bar"
                   created_at="1414770242000" updated_at="1414770242000" />
 
 </dataset>
diff --git a/sonar-core/src/test/resources/org/sonar/core/source/db/FileSourceDaoTest/shared.xml b/sonar-core/src/test/resources/org/sonar/core/source/db/FileSourceDaoTest/shared.xml
index fb3c258ddc0..538240d2fac 100644
--- a/sonar-core/src/test/resources/org/sonar/core/source/db/FileSourceDaoTest/shared.xml
+++ b/sonar-core/src/test/resources/org/sonar/core/source/db/FileSourceDaoTest/shared.xml
@@ -2,6 +2,7 @@
 
     <file_sources id="101" project_uuid="abcd" file_uuid="ab12"
                   data="aef12a,alice,2014-04-25T12:34:56+0100,,class Foo" data_hash="hash"
+                  line_hashes="truc"
                   created_at="1414597442000" updated_at="1414683842000" />
 
 </dataset>
diff --git a/sonar-core/src/test/resources/org/sonar/core/source/db/FileSourceDaoTest/update-result.xml b/sonar-core/src/test/resources/org/sonar/core/source/db/FileSourceDaoTest/update-result.xml
index 465b2e52cb8..ddad2e2b410 100644
--- a/sonar-core/src/test/resources/org/sonar/core/source/db/FileSourceDaoTest/update-result.xml
+++ b/sonar-core/src/test/resources/org/sonar/core/source/db/FileSourceDaoTest/update-result.xml
@@ -2,6 +2,7 @@
 
     <file_sources id="101" project_uuid="abcd" file_uuid="ab12"
                   data="updated data" data_hash="hash2"
+                  line_hashes="foo2&#10;bar2"
                   created_at="1414597442000" updated_at="1414770242000" />
 
 
diff --git a/sonar-plugin-api/src/main/java/org/sonar/api/batch/fs/internal/DefaultInputFile.java b/sonar-plugin-api/src/main/java/org/sonar/api/batch/fs/internal/DefaultInputFile.java
index a328071b25a..24f965709bc 100644
--- a/sonar-plugin-api/src/main/java/org/sonar/api/batch/fs/internal/DefaultInputFile.java
+++ b/sonar-plugin-api/src/main/java/org/sonar/api/batch/fs/internal/DefaultInputFile.java
@@ -40,6 +40,7 @@ public class DefaultInputFile implements InputFile, Serializable {
   private int lines;
   private String encoding;
   long[] originalLineOffsets;
+  String[] lineHashes;
 
   public DefaultInputFile(String moduleKey, String relativePath) {
     this.moduleKey = moduleKey;
@@ -113,6 +114,10 @@ public class DefaultInputFile implements InputFile, Serializable {
     return originalLineOffsets;
   }
 
+  public String[] lineHashes() {
+    return lineHashes;
+  }
+
   public DefaultInputFile setAbsolutePath(String s) {
     this.absolutePath = PathUtils.sanitize(s);
     return this;
@@ -158,6 +163,11 @@ public class DefaultInputFile implements InputFile, Serializable {
     return this;
   }
 
+  public DefaultInputFile setLineHashes(String[] lineHashes) {
+    this.lineHashes = lineHashes;
+    return this;
+  }
+
   @Override
   public boolean equals(Object o) {
     if (this == o) {