/**
* Import the source code corpus from the given directory
*
* @param srcDirPath
* @return
* @throws Exception
*/
public static SourceCodeCorpus importCodeData() throws Exception {
String srcDirPath = Config.getInstance().getCodeCorpusDir();
SourceCodeCorpus corpus = new SourceCodeCorpus();
File srcDir = new File(srcDirPath);
if (!srcDir.isDirectory()) {
System.out.println("The input directory path is invalid!");
return corpus;
}
// read the basic information from the "/basicInfo"
String basicInfoFilePath = Paths.get(srcDirPath, "basicInfo").toString();
if (!new File(basicInfoFilePath).isFile()) {
System.out.println("The file recording the basic information is missing!");
} else {
BufferedReader basicInfoReader = new BufferedReader(new FileReader(basicInfoFilePath));
// read the first line: fileType=\tfileType
String[] strs = basicInfoReader.readLine().split("\t");
String fileType = strs[1].trim();
// read the second line: segmentationLength=\tsegmentationLength
strs = basicInfoReader.readLine().split("\t");
int segmentationLength = Integer.parseInt(strs[1].trim());
corpus.setFileType(fileType);
corpus.setSegmentationLength(segmentationLength);
basicInfoReader.close();
}
// read the code content information from the "/codeContentCorpus"
String codeContentDirPath = Paths.get(srcDirPath, "codeContentCorpus").toString();
File codeContentDir = new File(codeContentDirPath);
if (!codeContentDir.isDirectory()) {
System.out.println("The directory for original code contents is missing!");
} else {
for (File oneFile : codeContentDir.listFiles()) {
if (oneFile.isFile()) {
String fullClassName = oneFile.getName();
BufferedReader reader = new BufferedReader(new FileReader(oneFile));
String codeContent = reader.readLine();
// If exists the full class name, set the code content;
// else create a new SourceCode object and set the code consent
boolean isExist = false;
for (SourceCode sourceCode : corpus.getSourceCodeList()) {
if (sourceCode.getFullClassName().equals(fullClassName.trim())) {
sourceCode.setContent(codeContent);
isExist = true;
break;
}
}
if (!isExist) {
SourceCode newSourceCode = new SourceCode();
newSourceCode.setFullClassName(fullClassName);
newSourceCode.setContent(codeContent);
corpus.addSourceCode(newSourceCode);
}
reader.close();
}
}
}
// read the code segment information from code segment corpus
String codeSegmentDirPath = Paths.get(srcDirPath, "codeSegmentCorpus").toString();
File codeSegmentDir = new File(codeSegmentDirPath);
if (!codeSegmentDir.isDirectory()) {
System.out.println("The directory for code segments is missing!");
} else {
for (File oneFile : codeContentDir.listFiles()) {
if (oneFile.isFile()) {
// fullClassName+@+"segment index".java
String oneSegmentName = oneFile.getName();
BufferedReader reader = new BufferedReader(new FileReader(oneFile));
String codeSegment = reader.readLine();
String[] strs = oneSegmentName.split("@");
String fullClassName = strs[0].trim();
// If exists the full class name, add the segment to the list;
// else create a new SourceCode object and add the segment to the list
boolean isExist = false;
for (SourceCode sourceCode : corpus.getSourceCodeList()) {
if (sourceCode.getFullClassName().equals(fullClassName.trim())) {
if (!sourceCode.getCodeSegmentList().contains(codeSegment)) {
sourceCode.addCodeSegment(codeSegment);
}
isExist = true;
break;
}
}
if (!isExist) {
SourceCode newSourceCode = new SourceCode();
newSourceCode.setFullClassName(fullClassName);
newSourceCode.addCodeSegment(codeSegment);
corpus.addSourceCode(newSourceCode);
}
reader.close();
}
}
}
// read the class names information from the "/classNameCorpus"
String classNameDirPath = Paths.get(srcDirPath, "classNameCorpus").toString();
File classNameDir = new File(classNameDirPath);
if (!classNameDir.isDirectory()) {
System.out.println("The directory for class names is missing!");
} else {
for (File oneFile : classNameDir.listFiles()) {
if (oneFile.isFile()) {
String fullClassName = oneFile.getName();
BufferedReader reader = new BufferedReader(new FileReader(oneFile));
String classNamesString = reader.readLine();
// If exists the full class name, add the class names to the list;
// else create a new SourceCode object and add all the class names to the list
boolean isExist = false;
for (SourceCode sourceCode : corpus.getSourceCodeList()) {
if (sourceCode.getFullClassName().equals(fullClassName.trim())) {
if (classNamesString == null) {
isExist = true;
break;
}
for (String oneClassName : classNamesString.split(" ")) {
if (!sourceCode.getClassNameList().contains(oneClassName.trim())) {
sourceCode.addClassName(oneClassName.trim());
}
}
isExist = true;
break;
}
}
if (!isExist) {
SourceCode newSourceCode = new SourceCode();
newSourceCode.setFullClassName(fullClassName);
if (classNamesString != null) {
for (String oneClassName : classNamesString.split(" ")) {
newSourceCode.addClassName(oneClassName.trim());
}
}
corpus.addSourceCode(newSourceCode);
}
reader.close();
}
}
}
// read the method names information from the "/methodNameCorpus"
String methodNameDirPath = Paths.get(srcDirPath, "methodNameCorpus").toString();
File methodNameDir = new File(methodNameDirPath);
if (!methodNameDir.isDirectory()) {
System.out.println("The directory for method names is missing!");
} else {
for (File oneFile : methodNameDir.listFiles()) {
if (oneFile.isFile()) {
String fullClassName = oneFile.getName();
BufferedReader reader = new BufferedReader(new FileReader(oneFile));
String methodNamesString = reader.readLine();
// If exists the full class name, add the class names to the list;
// else create a new SourceCode object and add all the method names to the list
boolean isExist = false;
for (SourceCode sourceCode : corpus.getSourceCodeList()) {
if (sourceCode.getFullClassName().equals(fullClassName.trim())) {
if (methodNamesString == null) {
isExist = true;
break;
}
for (String oneMethodName : methodNamesString.split(" ")) {
if (!sourceCode.getMethodNameList().contains(oneMethodName.trim()))
sourceCode.addMethodName(oneMethodName.trim());
}
isExist = true;
break;
}
}
if (!isExist) {
SourceCode newSourceCode = new SourceCode();
newSourceCode.setFullClassName(fullClassName);
if (methodNamesString != null) {
for (String oneMethodName : methodNamesString.split(" ")) {
newSourceCode.addMethodName(oneMethodName.trim());
}
}
corpus.addSourceCode(newSourceCode);
}
reader.close();
}
}
}
return corpus;
}
