This commit is contained in:
Kui LIU
2017-08-05 11:25:36 +02:00
parent 3abf34df1f
commit 21d55e82da
11 changed files with 156 additions and 72 deletions
+4
View File
@@ -12,6 +12,10 @@ OUTPUT/
*.pdf
Dataset/
GumTreeInput/
GumTreeInput1/
GumTreeOutput1/
GumTreeResults/
*TestParser.java
src/main/java/edu/lu/uni/serval/FixPatternParser/TestParser.java
# Package Files #
+2 -2
View File
@@ -16,7 +16,7 @@
</properties>
<dependencies>
<dependency>
<!-- <dependency>
<groupId>edu.lu.uni.serval</groupId>
<artifactId>MyCluster</artifactId>
<version>0.0.1-SNAPSHOT</version>
@@ -26,7 +26,7 @@
<groupId>edu.lu.uni.serval</groupId>
<artifactId>MyFeatureLearner</artifactId>
<version>0.0.1-SNAPSHOT</version>
</dependency>
</dependency> -->
<dependency>
<groupId>edu.lu.uni</groupId>
@@ -8,7 +8,7 @@ public class RunnableParser implements Runnable {
private File revFile;
private File diffentryFile;
private Parser parser;
public RunnableParser(File prevFile, File revFile, File diffentryFile, Parser parser) {
this.prevFile = prevFile;
this.revFile = revFile;
@@ -36,9 +36,9 @@ public class CommitPatchHunkParser extends CommitPatchParser {
List<DiffEntryHunk> diffentryHunks = new DiffEntryReader().readHunks(diffEntryFile);
//Filter out the modify actions, which are not in the DiffEntry hunks.
HunkActionFilter hunkFilter = new HunkActionFilter();
List<HunkFixPattern> allHunkFixPatternss = hunkFilter.filterActionsByDiffEntryHunk2(diffentryHunks, actionSets, revFile, prevFile);
List<HunkFixPattern> allHunkFixPatterns = hunkFilter.filterActionsByDiffEntryHunk2(diffentryHunks, actionSets, revFile, prevFile);
for (HunkFixPattern hunkFixPattern : allHunkFixPatternss) {
for (HunkFixPattern hunkFixPattern : allHunkFixPatterns) {
// Range of buggy source code
int startLine = 0;
int endLine = 0;
@@ -37,6 +37,12 @@ public class CommitPatchSingleStatementParser extends CommitPatchParser {
if (actionSets.size() > 0) {
// DiffEntry Hunks: filter out big hunks.
List<DiffEntryHunk> diffentryHunks = new DiffEntryReader().readHunks(diffEntryFile);
CUCreator cuCreator = new CUCreator();
CompilationUnit prevUnit = cuCreator.createCompilationUnit(prevFile);
CompilationUnit revUnit = cuCreator.createCompilationUnit(revFile);
if (prevUnit == null || revUnit == null) {
return;
}
for (HierarchicalActionSet actionSet : actionSets) {
// position of buggy statements
int startPosition = 0;
@@ -80,31 +86,21 @@ public class CommitPatchSingleStatementParser extends CommitPatchParser {
endPosition2 = startPosition2 + newNode.getLength();
}
}
} else {// DEL actions and MOV actions: we don't need these
// actions, as for now.
} else {// DEL actions and MOV actions: we don't need these actions, as for now.
continue;
}
if (startPosition == 0 || startPosition2 == 0) {
continue;
}
CUCreator cuCreator = new CUCreator();
CompilationUnit prevUnit = cuCreator.createCompilationUnit(prevFile);
CompilationUnit revUnit = cuCreator.createCompilationUnit(revFile);
if (prevUnit == null || revUnit == null) {
continue;
}
// Get line numbers.
int startLine = prevUnit.getLineNumber(startPosition);
int endLine = prevUnit.getLineNumber(endPosition);
int startLine2 = revUnit.getLineNumber(startPosition2);
int endLine2 = revUnit.getLineNumber(endPosition2);
if (endLine - startLine >= Configuration.HUNK_SIZE - 2
|| endLine2 - startLine2 >= Configuration.HUNK_SIZE - 2)
if (endLine - startLine >= Configuration.HUNK_SIZE - 2 || endLine2 - startLine2 >= Configuration.HUNK_SIZE - 2)
continue;
// Filter out the modify actions, which are not in the DiffEntry
// hunks.
// Filter out the modify actions, which are not in the DiffEntry hunks.
DiffEntryHunk hunk = matchHunk(startLine, endLine, startLine2, endLine2, actionStr, diffentryHunks);
if (hunk == null) {
continue;
@@ -117,8 +113,7 @@ public class CommitPatchSingleStatementParser extends CommitPatchParser {
SimplifyTree abstractIdentifier = new SimplifyTree();
abstractIdentifier.abstractTree(actionSet);
SimpleTree simpleTree = actionSet.getSimpleTree();
if (simpleTree == null) { // Failed to get the simple tree for
// INS actions.
if (simpleTree == null) { // Failed to get the simple tree for INS actions.
continue;
}
@@ -142,12 +137,12 @@ public class CommitPatchSingleStatementParser extends CommitPatchParser {
this.patchesSourceCode += Configuration.PATCH_SIGNAL + "\n" + patchSourceCode + "\n";
this.sizes += size + "\n";
this.astEditScripts += astEditScripts + "\n";
// 2. source code: raw tokens
String rawTokenEditScripts = getRawTokenEditScripts(actionSet);
// 3. abstract identifiers:
String abstractIdentifiersEditScripts = getAbstractIdentifiersEditScripts(actionSet);
// 4. semi-source code:
String semiSourceCodeEditScripts = getSemiSourceCodeEditScripts(actionSet);
// // 2. source code: raw tokens
// String rawTokenEditScripts = getRawTokenEditScripts(actionSet);
// // 3. abstract identifiers:
// String abstractIdentifiersEditScripts = getAbstractIdentifiersEditScripts(actionSet);
// // 4. semi-source code:
// String semiSourceCodeEditScripts = getSemiSourceCodeEditScripts(actionSet);
// this.buggyTrees += Configuration.BUGGY_TREE_TOKEN + "\n" +
// simpleTree.toString() + "\n";
@@ -12,7 +12,6 @@ import edu.lu.uni.serval.FixPatternParser.Tokenizer;
import edu.lu.uni.serval.config.Configuration;
import edu.lu.uni.serval.diffentry.DiffEntryHunk;
import edu.lu.uni.serval.diffentry.DiffEntryReader;
import edu.lu.uni.serval.gumtree.regroup.ActionFilter;
import edu.lu.uni.serval.gumtree.regroup.HierarchicalActionSet;
import edu.lu.uni.serval.gumtree.regroup.HunkActionFilter;
import edu.lu.uni.serval.gumtree.regroup.HunkFixPattern;
@@ -21,7 +20,7 @@ import edu.lu.uni.serval.gumtree.regroup.SimplifyTree;
import edu.lu.uni.serval.utils.MapSorter;
/**
* Parse fix patterns with GumTree.
* Parse fix violations with GumTree.
*
* @author kui.liu
*
@@ -29,36 +28,44 @@ import edu.lu.uni.serval.utils.MapSorter;
public class FixedViolationHunkParser extends FixedViolationParser {
@Override
public void parseFixPatterns(File prevFile, File revFile, File positionsFile) {
public void parseFixPatterns(File prevFile, File revFile, File diffentryFile) {
// GumTree results
List<HierarchicalActionSet> actionSets = parseChangedSourceCodeWithGumTree(prevFile, revFile);
if (actionSets.size() > 0) {
Map<Integer, Integer> positions = readPositions(positionsFile);
Map<Integer, Integer> positions = readPositions();
if (positions.size() > 1) {
MapSorter<Integer, Integer> sorter = new MapSorter<>();
positions = sorter.sortByKeyAscending(positions);
}
List<DiffEntryHunk> diffentryHunks1 = new DiffEntryReader().readHunks(diffentryFile);
int index = 0;
int hunkListSize = diffentryHunks1.size();
// Map<Integer, DiffEntryHunk> diffentryHunks = new HashMap<>();
// Select hunks by positions of violations.
List<DiffEntryHunk> diffentryHunks = new ArrayList<>();
for (Map.Entry<Integer, Integer> entry : positions.entrySet()) {
// only statements
for (HierarchicalActionSet actionSet : actionSets) {
String astNodeType = actionSet.getAstNodeType();
if (astNodeType.endsWith("Statement") || "FieldDeclaration".equals(astNodeType)) {
}
int startRange = entry.getKey();
int endRange = entry.getValue();
for (; index < hunkListSize; index ++) {
DiffEntryHunk hunk = diffentryHunks1.get(index);
int startLine = hunk.getBugLineStartNum();
int range = hunk.getBugRange();
if (startRange > startLine + range) continue;
if (endRange < startLine) break;
// startRange and endRange
// diffentryHunks.put(startRange, hunk);
diffentryHunks.add(hunk);
}
}
ActionFilter filter = new ActionFilter();
// DiffEntry size: filter out big hunks.
List<DiffEntryHunk> diffentryHunks = new DiffEntryReader().readHunks(positionsFile);
//Filter out the modify actions, which are not in the DiffEntry hunks.
HunkActionFilter hunkFilter = new HunkActionFilter();
List<HunkFixPattern> allHunkFixPatternss = hunkFilter.filterActionsByDiffEntryHunk2(diffentryHunks, actionSets, revFile, prevFile);
for (HunkFixPattern hunkFixPattern : allHunkFixPatternss) {
List<HunkFixPattern> allHunkFixPatterns = hunkFilter.filterActionsByDiffEntryHunk2(diffentryHunks, actionSets, revFile, prevFile);
for (HunkFixPattern hunkFixPattern : allHunkFixPatterns) {
// Range of buggy source code
int startLine = 0;
int endLine = 0;
@@ -18,8 +18,14 @@ import edu.lu.uni.serval.utils.FileHelper;
*/
public class FixedViolationParser extends Parser {
File positionFile = null;
public void setPositionFile(File positionFile) {
this.positionFile = positionFile;
}
@Override
public void parseFixPatterns(File prevFile, File revFile, File positionFile) {
public void parseFixPatterns(File prevFile, File revFile, File diffentryFile) {
}
protected boolean inPositions(int startLine, int endLine, Map<Integer, Integer> positions) {
@@ -31,7 +37,7 @@ public class FixedViolationParser extends Parser {
return false;
}
protected Map<Integer, Integer> readPositions(File positionFile) {
protected Map<Integer, Integer> readPositions() {
Map<Integer, Integer> positions = new HashMap<>();
String fileContent = FileHelper.readFile(positionFile);
BufferedReader reader = null;
@@ -19,7 +19,7 @@ import edu.lu.uni.serval.gumtree.regroup.SimpleTree;
import edu.lu.uni.serval.gumtree.regroup.SimplifyTree;
/**
* Parse fix patterns with GumTree.
* Parse fixed violations with GumTree.
*
* @author kui.liu
*
@@ -27,13 +27,20 @@ import edu.lu.uni.serval.gumtree.regroup.SimplifyTree;
public class FixedViolationSingleStatementParser extends FixedViolationParser {
@Override
public void parseFixPatterns(File prevFile, File revFile, File positionFile) {
public void parseFixPatterns(File prevFile, File revFile, File diffentryFile) {
// GumTree results
List<HierarchicalActionSet> actionSets = parseChangedSourceCodeWithGumTree(prevFile, revFile);
if (actionSets.size() > 0) {
// DiffEntry Hunks: filter out big hunks.
Map<Integer, Integer> positions = readPositions(positionFile);
CUCreator cuCreator = new CUCreator();
CompilationUnit prevUnit = cuCreator.createCompilationUnit(prevFile);
CompilationUnit revUnit = cuCreator.createCompilationUnit(revFile);
if (prevUnit == null || revUnit == null) {
return;
}
// Read the positions of checked violations
Map<Integer, Integer> positions = readPositions();
for (HierarchicalActionSet actionSet : actionSets) {
// position of buggy statements
int startPosition = 0;
@@ -76,29 +83,13 @@ public class FixedViolationSingleStatementParser extends FixedViolationParser {
endPosition2 = startPosition2 + newNode.getLength();
}
}
} else if (actionStr.startsWith("MOV")) {// DEL actions and MOV actions: we don't need these actions, as for now.
continue;
// startPosition = actionSet.getStartPosition();
// endPosition = startPosition + actionSet.getLength();
// ITree node = actionSet.getNode().getParent();
// startPosition2 = node.getPos();
// endPosition2 = startPosition2 + node.getLength();
} else {
startPosition = actionSet.getStartPosition();
endPosition = startPosition + actionSet.getLength();
} else {// DEL actions and MOV actions: we don't need these actions, as for now.
continue;
}
if (startPosition == 0 || startPosition2 == 0) {
continue;
}
CUCreator cuCreator = new CUCreator();
CompilationUnit prevUnit = cuCreator.createCompilationUnit(prevFile);
CompilationUnit revUnit = cuCreator.createCompilationUnit(revFile);
if (prevUnit == null || revUnit == null) {
continue;
}
// Get line numbers.
int startLine = prevUnit.getLineNumber(startPosition);
int endLine = prevUnit.getLineNumber(endPosition);
@@ -138,11 +129,11 @@ public class FixedViolationSingleStatementParser extends FixedViolationParser {
this.sizes += size + "\n";
this.astEditScripts += astEditScripts + "\n";
// 2. source code: raw tokens
String rawTokenEditScripts = getRawTokenEditScripts(actionSet);
// 3. abstract identifiers:
String abstractIdentifiersEditScripts = getAbstractIdentifiersEditScripts(actionSet);
// 4. semi-source code:
String semiSourceCodeEditScripts = getSemiSourceCodeEditScripts(actionSet);
// String rawTokenEditScripts = getRawTokenEditScripts(actionSet);
// // 3. abstract identifiers:
// String abstractIdentifiersEditScripts = getAbstractIdentifiersEditScripts(actionSet);
// // 4. semi-source code:
// String semiSourceCodeEditScripts = getSemiSourceCodeEditScripts(actionSet);
// this.buggyTrees += Configuration.BUGGY_TREE_TOKEN + "\n" + simpleTree.toString() + "\n";
this.tokensOfSourceCode += Tokenizer.getTokensDeepFirst(simpleTree).trim() + "\n";
@@ -55,10 +55,10 @@ public class AkkaParser {
List<MessageFile> msgFiles = new ArrayList<>();
for (File file : files) {
if (!file.isDirectory()) continue;
String projectFolder = file.getPath();
File revFileFolder = new File(projectFolder + "/revFiles/");// revised file folder
File[] revFiles = revFileFolder.listFiles();
for (File revFile : revFiles) {
if (revFile.getName().endsWith(".java")) {
File prevFile = new File(projectFolder + "/prevFiles/prev_" + revFile.getName());// previous file
@@ -7,6 +7,7 @@ public class MessageFile {
private File revFile;
private File prevFile;
private File diffEntryFile;
private File positionFile;
public MessageFile(File revFile, File prevFile, File diffEntryFile) {
super();
@@ -26,5 +27,13 @@ public class MessageFile {
public File getDiffEntryFile() {
return diffEntryFile;
}
public File getPositionFile() {
return positionFile;
}
public void setPositionFile(File positionFile) {
this.positionFile = positionFile;
}
}
@@ -53,7 +53,7 @@ public class DiffEntryReader {
}
hunk.append(line + "\n");
}
if ((range < 7 && range2 < 7) || range == 0 || range2 == 0) { // filter out big hunks
DiffEntryHunk diffEntryHunk = new DiffEntryHunk(startLine, startLine2, range, range2);
diffEntryHunk.setHunk(hunk.toString());
@@ -77,4 +77,76 @@ public class DiffEntryReader {
return diffentryHunks;
}
/**
* Read all hunks without considering their sizes.
*
* @param diffentryFile
* @return
*/
public List<DiffEntryHunk> readHunks2(File diffentryFile) {
List<DiffEntryHunk> diffentryHunks = new ArrayList<>();
String content = FileHelper.readFile(diffentryFile);
BufferedReader reader = null;
try {
reader = new BufferedReader(new StringReader(content));
String line = null;
int startLine = 0;
int range = 0;
int startLine2 = 0;
int range2 = 0;
StringBuilder hunk = new StringBuilder();
while ((line = reader.readLine()) != null) {
if (RegExp.filterSignal(line.trim())) {
if (hunk.length() > 0) {
if (startLine > 0) {
DiffEntryHunk diffEntryHunk = new DiffEntryHunk(startLine, startLine2, range, range2);
diffEntryHunk.setHunk(hunk.toString());
diffentryHunks.add(diffEntryHunk);
}
hunk.setLength(0);
}
int plusIndex = line.indexOf("+");
String lineNum = line.substring(4, plusIndex);
String[] nums = lineNum.split(",");
startLine = Integer.parseInt(nums[0].trim());
if (nums.length == 2) {
range = Integer.parseInt(nums[1].trim());
}
String lineNum2 = line.substring(plusIndex) .trim();
lineNum2 = lineNum2.substring(1, lineNum2.length() - 2);
String[] nums2 = lineNum2.split(",");
startLine2 = Integer.parseInt(nums2[0].trim());
if (nums2.length == 2) {
range2 = Integer.parseInt(nums2[1].trim());
}
continue;
}
hunk.append(line + "\n");
}
if (startLine > 0) {
DiffEntryHunk diffEntryHunk = new DiffEntryHunk(startLine, startLine2, range, range2);
diffEntryHunk.setHunk(hunk.toString());
diffentryHunks.add(diffEntryHunk);
}
hunk.setLength(0);
} catch (IOException e) {
e.printStackTrace();
} finally {
try {
if (reader != null) {
reader.close();
reader = null;
}
} catch (IOException e) {
e.printStackTrace();
}
}
return diffentryHunks;
}
}