FileDecompressor.js 7.1 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237
  1. const fs = require('fs-extra');
  2. const zlib = require('zlib');
  3. const crypto = require('crypto');
  4. const path = require('path');
  5. const unbzip2Stream = require('unbzip2-stream');
  6. const tar = require('tar-fs');
  7. const DecompressZip = require('decompress-zip');
  8. const utils = require('./utils');
  9. const FileDetector = require('./FileDetector');
  10. class FileDecompressor {
  11. constructor() {
  12. this.detector = new FileDetector();
  13. }
  14. async decompressNested(filename, outputDir) {
  15. await fs.ensureDir(outputDir);
  16. const fileType = await this.detector.detectFile(filename);
  17. let result = {
  18. sourceFile: filename,
  19. sourceFileType: fileType,
  20. selectedFile: filename,
  21. filesDir: outputDir,
  22. files: []
  23. };
  24. if (!fileType || !(fileType.ext == 'zip' || fileType.ext == 'bz2' || fileType.ext == 'gz' || fileType.ext == 'tar')) {
  25. return result;
  26. }
  27. result.files = await this.decompressTarZZ(fileType.ext, filename, outputDir);
  28. let sel = filename;
  29. let max = 0;
  30. if (result.files.length) {
  31. //ищем файл с максимальным размером
  32. for (let file of result.files) {
  33. if (file.size > max) {
  34. sel = `${outputDir}/${file.path}`;
  35. max = file.size;
  36. }
  37. }
  38. }
  39. result.selectedFile = sel;
  40. if (sel != filename) {
  41. result.nesting = await this.decompressNested(sel, `${outputDir}/${utils.randomHexString(10)}`);
  42. }
  43. return result;
  44. }
  45. async unpack(filename, outputDir) {
  46. const fileType = await this.detector.detectFile(filename);
  47. if (!fileType)
  48. throw new Error('Не удалось определить формат файла');
  49. return await this.decompress(fileType.ext, filename, outputDir);
  50. }
  51. async unpackTarZZ(filename, outputDir) {
  52. const fileType = await this.detector.detectFile(filename);
  53. if (!fileType)
  54. throw new Error('Не удалось определить формат файла');
  55. return await this.decompressTarZZ(fileType.ext, filename, outputDir);
  56. }
  57. async decompressTarZZ(fileExt, filename, outputDir) {
  58. const files = await this.decompress(fileExt, filename, outputDir);
  59. if (fileExt == 'tar' || files.length != 1)
  60. return files;
  61. const tarFilename = `${outputDir}/${files[0].path}`;
  62. const fileType = await this.detector.detectFile(tarFilename);
  63. if (!fileType || fileType.ext != 'tar')
  64. return files;
  65. const newTarFilename = `${outputDir}/${utils.randomHexString(30)}`;
  66. await fs.rename(tarFilename, newTarFilename);
  67. const tarFiles = await this.decompress('tar', newTarFilename, outputDir);
  68. await fs.remove(newTarFilename);
  69. return tarFiles;
  70. }
  71. async decompress(fileExt, filename, outputDir) {
  72. let files = [];
  73. switch (fileExt) {
  74. case 'zip':
  75. files = await this.unZip(filename, outputDir);
  76. break;
  77. case 'bz2':
  78. files = await this.unBz2(filename, outputDir);
  79. break;
  80. case 'gz':
  81. files = await this.unGz(filename, outputDir);
  82. break;
  83. case 'tar':
  84. files = await this.unTar(filename, outputDir);
  85. break;
  86. default:
  87. throw new Error(`FileDecompressor: неизвестный формат файла '${fileExt}'`);
  88. }
  89. return files;
  90. }
  91. async unZip(filename, outputDir) {
  92. return new Promise((resolve, reject) => {
  93. const files = [];
  94. const unzipper = new DecompressZip(filename);
  95. unzipper.on('error', function(err) {
  96. reject(err);
  97. });
  98. unzipper.on('extract', function() {
  99. resolve(files);
  100. });
  101. unzipper.extract({
  102. path: outputDir,
  103. filter: function(file) {
  104. if (file.type == 'File')
  105. files.push({path: file.path, size: file.uncompressedSize});
  106. return true;
  107. }
  108. });
  109. });
  110. }
  111. unBz2(filename, outputDir) {
  112. return this.decompressByStream(unbzip2Stream(), filename, outputDir);
  113. }
  114. unGz(filename, outputDir) {
  115. return this.decompressByStream(zlib.createGunzip(), filename, outputDir);
  116. }
  117. unTar(filename, outputDir) {
  118. return new Promise((resolve, reject) => {
  119. const files = [];
  120. const tarExtract = tar.extract(outputDir, {
  121. map: (header) => {
  122. files.push({path: header.name, size: header.size});
  123. return header;
  124. }
  125. });
  126. tarExtract.on('finish', () => {
  127. resolve(files);
  128. });
  129. tarExtract.on('error', (err) => {
  130. reject(err);
  131. });
  132. const inputStream = fs.createReadStream(filename);
  133. inputStream.on('error', (err) => {
  134. reject(err);
  135. });
  136. inputStream.pipe(tarExtract);
  137. });
  138. }
  139. decompressByStream(stream, filename, outputDir) {
  140. return new Promise(async(resolve, reject) => {
  141. const file = {path: path.parse(filename).name};
  142. let outFilename = `${outputDir}/${file.path}`;
  143. if (await fs.pathExists(outFilename)) {
  144. file.path = `${utils.randomHexString(10)}-${file.path}`;
  145. outFilename = `${outputDir}/${file.path}`;
  146. }
  147. const inputStream = fs.createReadStream(filename);
  148. const outputStream = fs.createWriteStream(outFilename);
  149. outputStream.on('finish', async() => {
  150. try {
  151. file.size = (await fs.stat(outFilename)).size;
  152. } catch (e) {
  153. reject(e);
  154. }
  155. resolve([file]);
  156. });
  157. stream.on('error', (err) => {
  158. reject(err);
  159. });
  160. inputStream.on('error', (err) => {
  161. reject(err);
  162. });
  163. outputStream.on('error', (err) => {
  164. reject(err);
  165. });
  166. inputStream.pipe(stream).pipe(outputStream);
  167. });
  168. }
  169. async gzipBuffer(buf) {
  170. return new Promise((resolve, reject) => {
  171. zlib.gzip(buf, {level: 1}, (err, result) => {
  172. if (err) reject(err);
  173. resolve(result);
  174. });
  175. });
  176. }
  177. async gzipFileIfNotExists(filename, outDir) {
  178. const buf = await fs.readFile(filename);
  179. const hash = crypto.createHash('sha256').update(buf).digest('hex');
  180. const outFilename = `${outDir}/${hash}`;
  181. if (!await fs.pathExists(outFilename)) {
  182. await fs.writeFile(outFilename, await this.gzipBuffer(buf))
  183. } else {
  184. await utils.touchFile(outFilename);
  185. }
  186. return outFilename;
  187. }
  188. }
  189. module.exports = FileDecompressor;