SHOW:
|
|
- or go back to the newest paste.
| 1 | package task; | |
| 2 | ||
| 3 | import java.io.File; | |
| 4 | ||
| 5 | public interface Checker {
| |
| 6 | boolean check(File file); | |
| 7 | } | |
| 8 | ------------ | |
| 9 | import java.io.BufferedReader; | |
| 10 | import java.io.File; | |
| 11 | - | import java.nio.file.Paths; |
| 11 | + | |
| 12 | import java.io.FileNotFoundException; | |
| 13 | import java.io.IOException; | |
| 14 | import java.io.InputStreamReader; | |
| 15 | import java.io.PrintStream; | |
| 16 | import java.nio.file.FileSystems; | |
| 17 | import java.nio.file.Files; | |
| 18 | import java.nio.file.Path; | |
| 19 | import java.util.ArrayList; | |
| 20 | import java.util.Arrays; | |
| 21 | import java.util.Collections; | |
| 22 | import java.util.HashSet; | |
| 23 | - | import net.sf.jmimemagic.Magic; |
| 23 | + | |
| 24 | - | import net.sf.jmimemagic.MagicException; |
| 24 | + | |
| 25 | - | import net.sf.jmimemagic.MagicMatch; |
| 25 | + | |
| 26 | - | import net.sf.jmimemagic.MagicMatchNotFoundException; |
| 26 | + | |
| 27 | - | import net.sf.jmimemagic.MagicParseException; |
| 27 | + | |
| 28 | import java.util.concurrent.Executor; | |
| 29 | import java.util.concurrent.Executors; | |
| 30 | import java.util.concurrent.LinkedBlockingQueue; | |
| 31 | ||
| 32 | import com.ibm.icu.text.CharsetDetector; | |
| 33 | ||
| 34 | public class FileScan implements Runnable {
| |
| 35 | private String whatFind; | |
| 36 | private BlockingQueue<File> queue; | |
| 37 | private CountDownLatch latch; | |
| 38 | private File endOfWorkFile; | |
| 39 | ||
| 40 | public FileScan(String whatFind, BlockingQueue<File> queue, | |
| 41 | CountDownLatch latch, File endOfWorkFile) {
| |
| 42 | this.whatFind = whatFind; | |
| 43 | this.queue = queue; | |
| 44 | this.latch = latch; | |
| 45 | this.endOfWorkFile = endOfWorkFile; | |
| 46 | } | |
| 47 | ||
| 48 | public FileScan() {
| |
| 49 | } | |
| 50 | ||
| 51 | @Override | |
| 52 | public void run() {
| |
| 53 | ||
| 54 | while (true) {
| |
| 55 | try {
| |
| 56 | File file; | |
| 57 | file = queue.take(); | |
| 58 | ||
| 59 | if (file == endOfWorkFile) {
| |
| 60 | break; | |
| 61 | } | |
| 62 | ||
| 63 | scan(file); | |
| 64 | } catch (InterruptedException e) {
| |
| 65 | e.printStackTrace(); | |
| 66 | } | |
| 67 | } | |
| 68 | ||
| 69 | latch.countDown(); | |
| 70 | } | |
| 71 | ||
| 72 | private void scan(File file) {
| |
| 73 | Scanner scanner = null; | |
| 74 | int matches = 0; | |
| 75 | ||
| 76 | try {
| |
| 77 | scanner = new Scanner(file); | |
| 78 | } catch (FileNotFoundException e) {
| |
| 79 | System.out.println("File Not Found.");
| |
| 80 | e.printStackTrace(); | |
| 81 | } | |
| 82 | ||
| 83 | while (scanner.hasNext()) | |
| 84 | if (scanner.next().equals(whatFind)) {
| |
| 85 | matches++; | |
| 86 | } | |
| 87 | - | System.out.println(myStr); |
| 87 | + | |
| 88 | if (matches > 0) {
| |
| 89 | String myStr = String.format( | |
| 90 | "File: %s - and the number of matches " + "is: %d", | |
| 91 | - | public void askUserPathAndWord() {
|
| 91 | + | |
| 92 | PrintStream ps = new PrintStream(System.out, true); | |
| 93 | ps.println(myStr); | |
| 94 | } | |
| 95 | } | |
| 96 | ||
| 97 | private void askUserPathAndWord() {
| |
| 98 | try {
| |
| 99 | tryToAskUserPathAndWord(); | |
| 100 | } catch (IOException | RuntimeException e) {
| |
| 101 | System.out.println("Wrong input!");
| |
| 102 | e.printStackTrace(); | |
| 103 | } catch (InterruptedException e) {
| |
| 104 | System.out.println("Interrupted.");
| |
| 105 | e.printStackTrace(); | |
| 106 | } | |
| 107 | } | |
| 108 | ||
| 109 | private void tryToAskUserPathAndWord() throws IOException, | |
| 110 | InterruptedException {
| |
| 111 | PathAndWord pathAndWord = readPathAndWord(); | |
| 112 | ||
| 113 | if (pathAndWord.isProperlyInitialized()) {
| |
| 114 | performScan(pathAndWord, "GameOver.tmp"); | |
| 115 | System.out.println("Thank you!");
| |
| 116 | } else {
| |
| 117 | System.out.println("You did not enter anything");
| |
| 118 | } | |
| 119 | } | |
| 120 | ||
| 121 | private PathAndWord readPathAndWord() throws IOException {
| |
| 122 | System.out | |
| 123 | .println("Please, enter a Path and Word (which you want to find):");
| |
| 124 | ||
| 125 | BufferedReader bufferedReader = new BufferedReader( | |
| 126 | new InputStreamReader(System.in)); | |
| 127 | ||
| 128 | String path = readPath(bufferedReader); | |
| 129 | String whatFind = readWord(bufferedReader); | |
| 130 | return new PathAndWord(path, whatFind); | |
| 131 | } | |
| 132 | ||
| 133 | private String readPath(BufferedReader bufferedReader) throws IOException {
| |
| 134 | boolean ok = false; | |
| 135 | File pathInput; | |
| 136 | ||
| 137 | do {
| |
| 138 | System.out.println("Please enter a Path:");
| |
| 139 | pathInput = new File(bufferedReader.readLine()); | |
| 140 | if (pathInput.exists() && pathInput.isDirectory()) {
| |
| 141 | ok = true; | |
| 142 | } else {
| |
| 143 | System.out.println("Doesn't exist or isn't folder!");
| |
| 144 | } | |
| 145 | } while (!ok); | |
| 146 | ||
| 147 | return pathInput.getAbsolutePath(); | |
| 148 | } | |
| 149 | ||
| 150 | private String readWord(BufferedReader bufferedReader) throws IOException {
| |
| 151 | System.out.println("Please enter a Word:");
| |
| 152 | return bufferedReader.readLine(); | |
| 153 | } | |
| 154 | ||
| 155 | private void performScan(PathAndWord pathAndWord, String endOfWorkFileName) | |
| 156 | throws InterruptedException {
| |
| 157 | BlockingQueue<File> queue = new LinkedBlockingQueue<File>(); | |
| 158 | ||
| 159 | File endOfWorkFile = new File(endOfWorkFileName); | |
| 160 | CountDownLatch latch = new CountDownLatch(2); | |
| 161 | ||
| 162 | FolderScan folderScan = new FolderScan(pathAndWord.path, queue, latch, | |
| 163 | endOfWorkFile); | |
| 164 | FileScan fileScan = new FileScan(pathAndWord.whatFind, queue, latch, | |
| 165 | endOfWorkFile); | |
| 166 | ||
| 167 | Executor executor = Executors.newCachedThreadPool(); | |
| 168 | executor.execute(folderScan); | |
| 169 | executor.execute(fileScan); | |
| 170 | ||
| 171 | latch.await(); | |
| 172 | } | |
| 173 | ||
| 174 | /** | |
| 175 | * @param args | |
| 176 | */ | |
| 177 | ||
| 178 | public static void main(String[] args) {
| |
| 179 | long startTime = System.currentTimeMillis(); | |
| 180 | ||
| 181 | new FileScan().askUserPathAndWord(); | |
| 182 | ||
| 183 | long stopTime = System.currentTimeMillis(); | |
| 184 | long elapsedTime = stopTime - startTime; | |
| 185 | System.out.println("\nRuntime time " + elapsedTime + " milliseconds.");
| |
| 186 | } | |
| 187 | } | |
| 188 | ||
| 189 | class FolderScan implements Runnable {
| |
| 190 | ||
| 191 | FolderScan(String path, BlockingQueue<File> queue, CountDownLatch latch, | |
| 192 | File endOfWorkFile) {
| |
| 193 | - | new ExtentionCheckers(), /*new ProbeContentTypeCheker(),*/ |
| 193 | + | |
| 194 | - | new EncodingsCheck() )); |
| 194 | + | |
| 195 | this.latch = latch; | |
| 196 | this.endOfWorkFile = endOfWorkFile; | |
| 197 | - | public FolderScan() {
|
| 197 | + | |
| 198 | checkers = new ArrayList<Checker>(Arrays.asList( | |
| 199 | new ExtentionChecking(), new ProbeContentTypeCheking(), | |
| 200 | new EncodingChecking())); | |
| 201 | ||
| 202 | - | findFiles(path); |
| 202 | + | List<UncriticalChecker> checkers = Collections |
| 203 | - | queue.add(endOfWorkFile); |
| 203 | + | .singletonList(new UncriticalChecker()); |
| 204 | FolderScan folderScan = new FolderScan(path, queue, latch, | |
| 205 | endOfWorkFile, checkers); | |
| 206 | } | |
| 207 | ||
| 208 | public FolderScan(String path2, BlockingQueue<File> queue2, | |
| 209 | CountDownLatch latch2, File endOfWorkFile2, | |
| 210 | List<UncriticalChecker> checkers2) {
| |
| 211 | } | |
| 212 | ||
| 213 | - | boolean checksFailed = true; |
| 213 | + | |
| 214 | public void run() {
| |
| 215 | try {
| |
| 216 | findFiles(path); | |
| 217 | - | // if (currentFile.getAbsolutePath().equals( |
| 217 | + | queue.put(endOfWorkFile); |
| 218 | - | // "E:\\Document\\!Nazar\\BOOKS\\)\\PDF\\readme.txt")) {
|
| 218 | + | latch.countDown(); |
| 219 | - | // System.out.println(); |
| 219 | + | |
| 220 | - | // } |
| 220 | + | |
| 221 | } | |
| 222 | } | |
| 223 | - | checksFailed = false; |
| 223 | + | |
| 224 | private void findFiles(String path) {
| |
| 225 | ||
| 226 | try {
| |
| 227 | File root = new File(path); | |
| 228 | - | if (checksFailed) {
|
| 228 | + | |
| 229 | for (File currentFile : list) {
| |
| 230 | boolean checksPassed = true; | |
| 231 | if (currentFile.isDirectory()) {
| |
| 232 | findFiles(currentFile.getAbsolutePath()); | |
| 233 | } else {
| |
| 234 | for (Checker currentChecker : checkers) {
| |
| 235 | if (!currentChecker.check(currentFile)) {
| |
| 236 | checksPassed = false; | |
| 237 | break; | |
| 238 | } | |
| 239 | } | |
| 240 | ||
| 241 | if (checksPassed) {
| |
| 242 | queue.put(currentFile); | |
| 243 | } | |
| 244 | } | |
| 245 | } | |
| 246 | } catch (InterruptedException e) {
| |
| 247 | e.printStackTrace(); | |
| 248 | } | |
| 249 | ||
| 250 | } | |
| 251 | ||
| 252 | private String path; | |
| 253 | private BlockingQueue<File> queue; | |
| 254 | private CountDownLatch latch; | |
| 255 | private File endOfWorkFile; | |
| 256 | private List<Checker> checkers; | |
| 257 | private List<UncriticalChecker> checkList; | |
| 258 | } | |
| 259 | ||
| 260 | - | class ExtentionCheckers implements Checker {
|
| 260 | + | class UncriticalChecker implements Checker {
|
| 261 | @Override | |
| 262 | public boolean check(File currentFile) {
| |
| 263 | return true; | |
| 264 | } | |
| 265 | } | |
| 266 | ||
| 267 | class PathAndWord {
| |
| 268 | - | ".djv", ".rar", ".rtf")); |
| 268 | + | |
| 269 | final String whatFind; | |
| 270 | ||
| 271 | PathAndWord(String path, String whatFind) {
| |
| 272 | this.path = path; | |
| 273 | this.whatFind = whatFind; | |
| 274 | } | |
| 275 | ||
| 276 | boolean isProperlyInitialized() {
| |
| 277 | return path != null && whatFind != null; | |
| 278 | - | class EncodingsCheck implements Checker {
|
| 278 | + | |
| 279 | } | |
| 280 | ||
| 281 | class ExtentionChecking implements Checker {
| |
| 282 | - | return isUTF8(currentFile); |
| 282 | + | |
| 283 | @Override | |
| 284 | public boolean check(File currentFile) {
| |
| 285 | - | public static boolean isUTF8(File file) {
|
| 285 | + | |
| 286 | Set<String> extensions = new HashSet<String>(Arrays.asList(".txt",
| |
| 287 | ".pdf", ".doc", ".docx", ".html", ".htm", ".xml", ".djvu", | |
| 288 | ".djv", ".rar", ".rtf", ".tmp")); | |
| 289 | ||
| 290 | if (extensions.contains(fileName.substring(fileName.lastIndexOf(".")))) {
| |
| 291 | return true; | |
| 292 | } | |
| 293 | ||
| 294 | return false; | |
| 295 | } | |
| 296 | ||
| 297 | private String fileName; | |
| 298 | } | |
| 299 | ||
| 300 | class EncodingChecking implements Checker {
| |
| 301 | ||
| 302 | @Override | |
| 303 | public boolean check(File currentFile) {
| |
| 304 | - | if (0 == (buffer[0] & 0x80)) {
|
| 304 | + | return detectEncoding(currentFile); |
| 305 | - | return true; // ASCII subset character, fast path |
| 305 | + | |
| 306 | - | } else if (0xF0 == (buffer[0] & 0xF8)) { // start of 4-byte sequence
|
| 306 | + | |
| 307 | - | if (buffer[3] >= buffer.length) {
|
| 307 | + | public static boolean detectEncoding(File file) {
|
| 308 | - | return false; |
| 308 | + | detector = new CharsetDetector(); |
| 309 | ||
| 310 | - | if ((0x80 == (buffer[1] & 0xC0)) && (0x80 == (buffer[2] & 0xC0)) |
| 310 | + | |
| 311 | - | && (0x80 == (buffer[3] & 0xC0))) |
| 311 | + | |
| 312 | throw new IllegalArgumentException("input file can't be null");
| |
| 313 | - | } else if (0xE0 == (buffer[0] & 0xF0)) { // start of 3-byte sequence
|
| 313 | + | |
| 314 | - | if (buffer[2] >= buffer.length) {
|
| 314 | + | |
| 315 | - | return false; |
| 315 | + | |
| 316 | "input file refers to a directory"); | |
| 317 | - | if ((0x80 == (buffer[1] & 0xC0)) && (0x80 == (buffer[2] & 0xC0))) {
|
| 317 | + | |
| 318 | ||
| 319 | // read input file | |
| 320 | - | } else if (0xC0 == (buffer[0] & 0xE0)) { // start of 2-byte sequence
|
| 320 | + | |
| 321 | - | if (buffer[1] >= buffer.length) {
|
| 321 | + | |
| 322 | - | return false; |
| 322 | + | |
| 323 | } catch (IOException e) {
| |
| 324 | - | if (0x80 == (buffer[1] & 0xC0)) {
|
| 324 | + | |
| 325 | "Can't read input file, error = " + e.getLocalizedMessage()); | |
| 326 | } | |
| 327 | ||
| 328 | if (detector.setText(buffer) != null) {
| |
| 329 | return true; | |
| 330 | } | |
| 331 | ||
| 332 | return false; | |
| 333 | } | |
| 334 | ||
| 335 | - | try{
|
| 335 | + | |
| 336 | - | byte firstBytes[] = new byte[4]; |
| 336 | + | |
| 337 | - | int count = fileInputStream.read(firstBytes); |
| 337 | + | |
| 338 | - | if(count < 4){
|
| 338 | + | |
| 339 | - | throw new IOException("Empty file");
|
| 339 | + | byte firstBytes[] = new byte[50]; |
| 340 | - | } |
| 340 | + | int count = fileInputStream.read(firstBytes); |
| 341 | - | return firstBytes; |
| 341 | + | if (count < 5) {
|
| 342 | - | } finally {
|
| 342 | + | throw new IOException("Poor file!");
|
| 343 | - | fileInputStream.close(); |
| 343 | + | |
| 344 | - | } |
| 344 | + | return firstBytes; |
| 345 | } finally {
| |
| 346 | fileInputStream.close(); | |
| 347 | } | |
| 348 | - | class ProbeContentTypeCheker implements Checker {
|
| 348 | + | |
| 349 | ||
| 350 | private static CharsetDetector detector; | |
| 351 | } | |
| 352 | - | String mimeType = null; |
| 352 | + | |
| 353 | class ProbeContentTypeCheking implements Checker {
| |
| 354 | - | Path path = Paths.get(currentFile.getAbsolutePath()); |
| 354 | + | |
| 355 | - | byte[] data = Files.readAllBytes(path); |
| 355 | + | |
| 356 | - | MagicMatch match = Magic.getMagicMatch(data); |
| 356 | + | |
| 357 | - | mimeType = match.getMimeType(); |
| 357 | + | |
| 358 | - | } catch (MagicParseException | MagicMatchNotFoundException |
| 358 | + | Path filePath = FileSystems.getDefault().getPath( |
| 359 | - | | MagicException | IOException e) {
|
| 359 | + | currentFile.getAbsolutePath()); |
| 360 | if ((null != Files.probeContentType(filePath))) {
| |
| 361 | return true; | |
| 362 | } | |
| 363 | - | if (null != mimeType) {
|
| 363 | + | |
| 364 | e.printStackTrace(); | |
| 365 | } | |
| 366 | ||
| 367 | return false; | |
| 368 | } | |
| 369 | } |