/**
* @param graphDBNameOld
* @param graphDBName
* @throws IOException
* @throws ParseException
*/
public static void pruningUnconnected(String graphDBNameOld, String graphDBName)
throws IOException, ParseException {
System.out.println("In Pruning Unconnected");
SmilesParser sParser = new SmilesParser();
// InputStream file = new FileInputStream(graphDBNameOld);
// Graph[] graphs = sParser.parse(file, MyFactory.getGraphFactory());
BufferedReader bin = new BufferedReader(new FileReader(graphDBNameOld));
LinkedList<Graph> graphlists = new LinkedList<Graph>();
String line;
while ((line = bin.readLine()) != null) {
int pos = line.indexOf(" => ");
try {
graphlists.add(
sParser.parse(
line.substring(pos + " => ".length()),
line.substring(0, pos),
MyFactory.getGraphFactory()));
System.out.println("GraphID: " + line.substring(0, pos));
} catch (ParseException e) {
System.out.println(line);
e.printStackTrace();
}
}
Graph[] graphs = new Graph[graphlists.size()];
graphlists.toArray(graphs);
System.out.println("After Parsing all Graphs");
int connect = 0;
BufferedWriter indexWriter = new BufferedWriter(new FileWriter(graphDBName));
for (int i = 0; i < graphs.length; i++) {
if (GraphConnectivityTester.isConnected(graphs[i])) {
StringBuffer buf = new StringBuffer();
buf.append(connect);
buf.append(" => ");
buf.append(sParser.serialize(graphs[i]));
buf.append("\n");
indexWriter.write(buf.toString());
connect++;
}
}
indexWriter.close();
// Write the meta information of the smile data file:
BufferedWriter metaWriter = new BufferedWriter(new FileWriter(graphDBName + "_Meta"));
// 1. Processing Date
SimpleDateFormat bartDateFormat = new SimpleDateFormat("EEEE-MMMM-dd-yyyy");
Date date = new Date();
metaWriter.write(bartDateFormat.format(date));
metaWriter.newLine();
// 2. Number of graphs in this file
metaWriter.write("Number of Graphs:" + connect);
// Close meta data file
try {
metaWriter.close();
bin.close();
} catch (IOException e) {
e.printStackTrace();
}
System.out.println("TOTAL Graph: " + graphs.length + "Connected : " + connect);
}
public static void changeFormatSimple(
String SDFFileName, String SmilesFileName, GraphFactory graphFactory) throws IOException {
// Open InputFile
BufferedReader fileBufReader = new BufferedReader(new FileReader(SDFFileName));
// Open OutputFile
BufferedWriter outputWriter = new BufferedWriter(new FileWriter(SmilesFileName));
SDFParserModified SDFParser = MyFactory.getSDFParserM();
SmilesParser smilesParser = MyFactory.getSmilesParser();
Graph oneGraph = null;
double aveEdgeCount = 0;
double aveNodeCount = 0;
int index = 0;
String spliter = " => ";
while (fileBufReader.ready()) {
try {
oneGraph = SDFParser.parse(fileBufReader, graphFactory);
} catch (ParseException e) {
System.out.println("skip one graph");
while (fileBufReader.ready()) {
String aLine = fileBufReader.readLine();
if (aLine.equals("$$$$")) break;
}
continue;
}
if (GraphConnectivityTester.isConnected(oneGraph)) {
aveEdgeCount =
(aveEdgeCount) * (index / ((double) index + 1))
+ oneGraph.getEdgeCount() / ((double) index + 1);
aveNodeCount =
(aveNodeCount) * (index / ((double) index + 1))
+ oneGraph.getNodeCount() / ((double) index + 1);
if (index > 0) outputWriter.newLine();
outputWriter.write(index + spliter + smilesParser.serialize(oneGraph));
index++;
}
}
// Close input File
try {
fileBufReader.close();
outputWriter.flush();
outputWriter.close();
} catch (IOException e) {
e.printStackTrace();
}
System.out.println(
"In processor: changeFormat, "
+ index
+ " number of graphs"
+ "has been formated into Smiles Format");
// Intrigue java garbage collector
Runtime r = Runtime.getRuntime();
r.gc();
// Write the meta information of the smile data file:
BufferedWriter metaWriter = new BufferedWriter(new FileWriter(SmilesFileName + "_Meta"));
// 1. Processing Date
SimpleDateFormat bartDateFormat = new SimpleDateFormat("EEEE-MMMM-dd-yyyy");
Date date = new Date();
metaWriter.write(bartDateFormat.format(date));
metaWriter.newLine();
// 2. Number of graphs in this file
metaWriter.write("Number of Graphs:" + index);
metaWriter.write("Ave EdgeCount: " + aveEdgeCount + " Ave NodeCount: " + aveNodeCount);
// Close meta data file
try {
metaWriter.flush();
metaWriter.close();
} catch (IOException e) {
e.printStackTrace();
}
}
/**
* Given a file of graphs in SDF format Populate graphs in this file into A new graph file [with
* ID = > in Smiles format] Pay attention: all graphs have to be connected
*
* @param SDFFileName
* @param graphFactory
* @throws IOException
* @throws DataAccessException
* @throws IOException
*/
public static void changeFormat(
String SDFFileName, String SmilesFileName, GraphFactory graphFactory) throws IOException {
// Open InputFile
BufferedReader fileBufReader = new BufferedReader(new FileReader(SDFFileName));
// Open OutputFile
BufferedWriter outputWriter = new BufferedWriter(new FileWriter(SmilesFileName));
SDFParserModified SDFParser = MyFactory.getSDFParserM();
SmilesParser smilesParser = MyFactory.getSmilesParser();
String lineString;
StringBuffer graphBuffer = new StringBuffer(1024);
Graph oneGraph = null;
int index = 0;
String spliter = " => ";
boolean attStatus = false;
while ((lineString = fileBufReader.readLine()) != null) {
if (lineString.startsWith(">")) {
// enter the attribute status
attStatus = true;
}
if (attStatus) {
// One graph Has been read
if (attStatus && lineString.equals("$$$$")) {
// Transform this graphString to a realGraph and save into
// database
try {
oneGraph = SDFParser.parse(graphBuffer.toString(), graphFactory);
} catch (ParseException e) {
e.printStackTrace();
}
if (GraphConnectivityTester.isConnected(oneGraph)) {
if (index > 0) outputWriter.newLine();
outputWriter.write(index + spliter + smilesParser.serialize(oneGraph));
index++;
// System.out.println(index + ":" + oneGraph.getName());
}
graphBuffer.delete(0, graphBuffer.length());
attStatus = false;
} else continue;
} else graphBuffer = graphBuffer.append(lineString + '\n');
}
// Close input File
try {
fileBufReader.close();
} catch (IOException e) {
e.printStackTrace();
}
// Close out File
try {
outputWriter.newLine();
outputWriter.close();
} catch (IOException e) {
e.printStackTrace();
}
System.out.println(
"In processor: changeFormat, "
+ index
+ " number of graphs"
+ "has been formated into Smiles Format");
// Intrigue java garbage collector
Runtime r = Runtime.getRuntime();
r.gc();
// Write the meta information of the smile data file:
BufferedWriter metaWriter = new BufferedWriter(new FileWriter(SmilesFileName + "_Meta"));
// 1. Processing Date
SimpleDateFormat bartDateFormat = new SimpleDateFormat("EEEE-MMMM-dd-yyyy");
Date date = new Date();
metaWriter.write(bartDateFormat.format(date));
metaWriter.newLine();
// 2. Number of graphs in this file
metaWriter.write("Number of Graphs:" + index);
// Close meta data file
try {
metaWriter.close();
} catch (IOException e) {
e.printStackTrace();
}
}