130 lines
4.7 KiB
Java
130 lines
4.7 KiB
Java
package wspp;
|
|
|
|
import java.io.*;
|
|
import java.util.*;
|
|
|
|
/**
|
|
* @author Nikita Doschennikov (me@fymio.us)
|
|
*/
|
|
public class WsppPosition {
|
|
|
|
public static void main(String[] args) {
|
|
if (args.length != 2) {
|
|
System.err.println(
|
|
"usage: java WsppPosition <inputFilePath> <outputFilePath>"
|
|
);
|
|
}
|
|
|
|
final String inputFileName = args[0];
|
|
final String outputFileName = args[1];
|
|
|
|
Map<String, WordInfo> words = new LinkedHashMap<>();
|
|
|
|
try (
|
|
BufferedReader br = new BufferedReader(
|
|
new FileReader(inputFileName)
|
|
);
|
|
FileWriter fw = new FileWriter(outputFileName)
|
|
) {
|
|
String line;
|
|
int wordPos = 1;
|
|
int lineNumber = 1;
|
|
while ((line = br.readLine()) != null) {
|
|
line = line.toLowerCase();
|
|
StringBuilder word = new StringBuilder();
|
|
|
|
for (char c : line.toCharArray()) {
|
|
if (
|
|
Character.isLetter(c) ||
|
|
c == '\'' ||
|
|
Character.getType(c) == Character.DASH_PUNCTUATION ||
|
|
Character.isDigit(c) ||
|
|
c == '$' ||
|
|
c == '_'
|
|
) {
|
|
word.append(c);
|
|
} else {
|
|
if (!word.isEmpty()) {
|
|
if (words.containsKey(word.toString())) {
|
|
var lO = words.get(
|
|
word.toString()
|
|
).lineOccurrences;
|
|
if (lO.containsKey(lineNumber)) {
|
|
lO.get(lineNumber).put(wordPos++);
|
|
} else {
|
|
var intList = new IntList();
|
|
intList.put(wordPos++);
|
|
lO.put(lineNumber, intList);
|
|
}
|
|
} else {
|
|
WordInfo info = new WordInfo(
|
|
word.toString(),
|
|
wordPos
|
|
);
|
|
var intList = new IntList();
|
|
intList.put(wordPos++);
|
|
info.lineOccurrences.put(lineNumber, intList);
|
|
words.put(word.toString(), info);
|
|
}
|
|
}
|
|
|
|
word = new StringBuilder();
|
|
}
|
|
}
|
|
|
|
if (!word.isEmpty()) {
|
|
if (words.containsKey(word.toString())) {
|
|
var lO = words.get(word.toString()).lineOccurrences;
|
|
if (lO.containsKey(lineNumber)) {
|
|
lO.get(lineNumber).put(wordPos++);
|
|
} else {
|
|
var intList = new IntList();
|
|
intList.put(wordPos++);
|
|
lO.put(lineNumber, intList);
|
|
}
|
|
} else {
|
|
WordInfo info = new WordInfo(word.toString(), wordPos);
|
|
var intList = new IntList();
|
|
intList.put(wordPos++);
|
|
info.lineOccurrences.put(lineNumber, intList);
|
|
words.put(word.toString(), info);
|
|
}
|
|
}
|
|
|
|
lineNumber++;
|
|
}
|
|
|
|
List<WordInfo> sortedWords = new ArrayList<>(words.values());
|
|
sortedWords.sort(
|
|
Comparator.comparingInt((WordInfo w) ->
|
|
w.word.length()
|
|
).thenComparingInt(w -> w.firstOccurrence)
|
|
);
|
|
|
|
for (WordInfo info : sortedWords) {
|
|
int totalNumberOfOccurrences = 0;
|
|
var lO = info.lineOccurrences;
|
|
String word = info.word;
|
|
for (int key : lO.keySet()) {
|
|
totalNumberOfOccurrences += lO.get(key).getLength();
|
|
}
|
|
|
|
fw.write(word + " " + totalNumberOfOccurrences);
|
|
|
|
for (int key : lO.keySet()) {
|
|
var occurrences = lO.get(key);
|
|
|
|
for (int i = 0; i < occurrences.getLength(); i++) {
|
|
fw.write(
|
|
" " + key + ":" + (wordPos - occurrences.get(i))
|
|
);
|
|
}
|
|
}
|
|
fw.write("\n");
|
|
}
|
|
} catch (IOException e) {
|
|
System.out.println("Error reading file.");
|
|
}
|
|
}
|
|
}
|