@push.rocks/smartarchive 4.2.4 → 5.0.0

package/ts/classes.smartarchive.ts

@@ -1,75 +1,267 @@
 import * as plugins from './plugins.js';
-import * as paths from './paths.js';
+import type {
+  IArchiveCreationOptions,
+  IArchiveEntry,
+  IArchiveExtractionOptions,
+  IArchiveEntryInfo,
+  IArchiveInfo,
+  TArchiveFormat,
+  TCompressionLevel,
+} from './interfaces.js';
 
 import { Bzip2Tools } from './classes.bzip2tools.js';
 import { GzipTools } from './classes.gziptools.js';
 import { TarTools } from './classes.tartools.js';
 import { ZipTools } from './classes.ziptools.js';
+import { ArchiveAnalyzer, type IAnalyzedResult } from './classes.archiveanalyzer.js';
 
-import {
-  ArchiveAnalyzer,
-  type IAnalyzedResult,
-} from './classes.archiveanalyzer.js';
-
-import type { from } from '@push.rocks/smartrx/dist_ts/smartrx.plugins.rxjs.js';
-
+/**
+ * Main class for archive manipulation
+ * Supports TAR, ZIP, GZIP, and BZIP2 formats
+ */
 export class SmartArchive {
-  // STATIC
-  public static async fromArchiveUrl(urlArg: string): Promise<SmartArchive> {
+  // ============================================
+  // STATIC FACTORY METHODS - EXTRACTION
+  // ============================================
+
+  /**
+   * Create SmartArchive from a URL
+   */
+  public static async fromUrl(urlArg: string): Promise<SmartArchive> {
     const smartArchiveInstance = new SmartArchive();
     smartArchiveInstance.sourceUrl = urlArg;
     return smartArchiveInstance;
   }
 
-  public static async fromArchiveFile(
-    filePathArg: string,
-  ): Promise<SmartArchive> {
+  /**
+   * Create SmartArchive from a local file path
+   */
+  public static async fromFile(filePathArg: string): Promise<SmartArchive> {
     const smartArchiveInstance = new SmartArchive();
     smartArchiveInstance.sourceFilePath = filePathArg;
     return smartArchiveInstance;
   }
 
-  public static async fromArchiveStream(
-    streamArg:
-      | plugins.stream.Readable
-      | plugins.stream.Duplex
-      | plugins.stream.Transform,
+  /**
+   * Create SmartArchive from a readable stream
+   */
+  public static async fromStream(
+    streamArg: plugins.stream.Readable | plugins.stream.Duplex | plugins.stream.Transform
   ): Promise<SmartArchive> {
     const smartArchiveInstance = new SmartArchive();
     smartArchiveInstance.sourceStream = streamArg;
     return smartArchiveInstance;
   }
 
-  // INSTANCE
+  /**
+   * Create SmartArchive from an in-memory buffer
+   */
+  public static async fromBuffer(buffer: Buffer): Promise<SmartArchive> {
+    const smartArchiveInstance = new SmartArchive();
+    smartArchiveInstance.sourceStream = plugins.stream.Readable.from(buffer);
+    return smartArchiveInstance;
+  }
+
+  // ============================================
+  // STATIC FACTORY METHODS - CREATION
+  // ============================================
+
+  /**
+   * Create a new archive from a directory
+   */
+  public static async fromDirectory(
+    directoryPath: string,
+    options: IArchiveCreationOptions
+  ): Promise<SmartArchive> {
+    const smartArchiveInstance = new SmartArchive();
+    smartArchiveInstance.creationOptions = options;
+
+    const tarTools = new TarTools();
+
+    if (options.format === 'tar' || options.format === 'tar.gz' || options.format === 'tgz') {
+      if (options.format === 'tar') {
+        const pack = await tarTools.packDirectory(directoryPath);
+        pack.finalize();
+        smartArchiveInstance.archiveBuffer = await SmartArchive.streamToBuffer(pack);
+      } else {
+        smartArchiveInstance.archiveBuffer = await tarTools.packDirectoryToTarGz(
+          directoryPath,
+          options.compressionLevel
+        );
+      }
+    } else if (options.format === 'zip') {
+      const zipTools = new ZipTools();
+      const fileTree = await plugins.listFileTree(directoryPath, '**/*');
+      const entries: IArchiveEntry[] = [];
+
+      for (const filePath of fileTree) {
+        const absolutePath = plugins.path.join(directoryPath, filePath);
+        const content = await plugins.fsPromises.readFile(absolutePath);
+        entries.push({
+          archivePath: filePath,
+          content,
+        });
+      }
+
+      smartArchiveInstance.archiveBuffer = await zipTools.createZip(entries, options.compressionLevel);
+    } else {
+      throw new Error(`Unsupported format for directory packing: ${options.format}`);
+    }
+
+    return smartArchiveInstance;
+  }
+
+  /**
+   * Create a new archive from an array of entries
+   */
+  public static async fromFiles(
+    files: IArchiveEntry[],
+    options: IArchiveCreationOptions
+  ): Promise<SmartArchive> {
+    const smartArchiveInstance = new SmartArchive();
+    smartArchiveInstance.creationOptions = options;
+
+    if (options.format === 'tar' || options.format === 'tar.gz' || options.format === 'tgz') {
+      const tarTools = new TarTools();
+      if (options.format === 'tar') {
+        smartArchiveInstance.archiveBuffer = await tarTools.packFiles(files);
+      } else {
+        smartArchiveInstance.archiveBuffer = await tarTools.packFilesToTarGz(files, options.compressionLevel);
+      }
+    } else if (options.format === 'zip') {
+      const zipTools = new ZipTools();
+      smartArchiveInstance.archiveBuffer = await zipTools.createZip(files, options.compressionLevel);
+    } else if (options.format === 'gz') {
+      if (files.length !== 1) {
+        throw new Error('GZIP format only supports a single file');
+      }
+      const gzipTools = new GzipTools();
+      let content: Buffer;
+      if (typeof files[0].content === 'string') {
+        content = Buffer.from(files[0].content);
+      } else if (Buffer.isBuffer(files[0].content)) {
+        content = files[0].content;
+      } else {
+        throw new Error('GZIP format requires string or Buffer content');
+      }
+      smartArchiveInstance.archiveBuffer = await gzipTools.compress(content, options.compressionLevel);
+    } else {
+      throw new Error(`Unsupported format: ${options.format}`);
+    }
+
+    return smartArchiveInstance;
+  }
+
+  /**
+   * Start building an archive incrementally using a builder pattern
+   */
+  public static create(options: IArchiveCreationOptions): SmartArchive {
+    const smartArchiveInstance = new SmartArchive();
+    smartArchiveInstance.creationOptions = options;
+    smartArchiveInstance.pendingEntries = [];
+    return smartArchiveInstance;
+  }
+
+  /**
+   * Helper to convert a stream to buffer
+   */
+  private static async streamToBuffer(stream: plugins.stream.Readable): Promise<Buffer> {
+    const chunks: Buffer[] = [];
+    return new Promise((resolve, reject) => {
+      stream.on('data', (chunk) => chunks.push(Buffer.isBuffer(chunk) ? chunk : Buffer.from(chunk)));
+      stream.on('end', () => resolve(Buffer.concat(chunks)));
+      stream.on('error', reject);
+    });
+  }
+
+  // ============================================
+  // INSTANCE PROPERTIES
+  // ============================================
+
   public tarTools = new TarTools();
   public zipTools = new ZipTools();
   public gzipTools = new GzipTools();
   public bzip2Tools = new Bzip2Tools(this);
   public archiveAnalyzer = new ArchiveAnalyzer(this);
 
-  public sourceUrl: string;
-  public sourceFilePath: string;
-  public sourceStream:
-    | plugins.stream.Readable
-    | plugins.stream.Duplex
-    | plugins.stream.Transform;
+  public sourceUrl?: string;
+  public sourceFilePath?: string;
+  public sourceStream?: plugins.stream.Readable | plugins.stream.Duplex | plugins.stream.Transform;
 
-  public archiveName: string;
-  public singleFileMode: boolean = false;
-
-  public addedDirectories: string[] = [];
-  public addedFiles: (
-    | plugins.smartfile.SmartFile
-    | plugins.smartfile.StreamFile
-  )[] = [];
-  public addedUrls: string[] = [];
+  private archiveBuffer?: Buffer;
+  private creationOptions?: IArchiveCreationOptions;
+  private pendingEntries?: IArchiveEntry[];
 
   constructor() {}
 
+  // ============================================
+  // BUILDER METHODS (for incremental creation)
+  // ============================================
+
+  /**
+   * Add a file to the archive (builder pattern)
+   */
+  public addFile(archivePath: string, content: string | Buffer): this {
+    if (!this.pendingEntries) {
+      throw new Error('addFile can only be called on archives created with SmartArchive.create()');
+    }
+    this.pendingEntries.push({ archivePath, content });
+    return this;
+  }
+
+  /**
+   * Add a SmartFile to the archive (builder pattern)
+   */
+  public addSmartFile(file: plugins.smartfile.SmartFile, archivePath?: string): this {
+    if (!this.pendingEntries) {
+      throw new Error('addSmartFile can only be called on archives created with SmartArchive.create()');
+    }
+    this.pendingEntries.push({
+      archivePath: archivePath || file.relative,
+      content: file,
+    });
+    return this;
+  }
+
+  /**
+   * Add a StreamFile to the archive (builder pattern)
+   */
+  public addStreamFile(file: plugins.smartfile.StreamFile, archivePath?: string): this {
+    if (!this.pendingEntries) {
+      throw new Error('addStreamFile can only be called on archives created with SmartArchive.create()');
+    }
+    this.pendingEntries.push({
+      archivePath: archivePath || file.relativeFilePath,
+      content: file,
+    });
+    return this;
+  }
+
+  /**
+   * Build the archive from pending entries
+   */
+  public async build(): Promise<SmartArchive> {
+    if (!this.pendingEntries || !this.creationOptions) {
+      throw new Error('build can only be called on archives created with SmartArchive.create()');
+    }
+
+    const built = await SmartArchive.fromFiles(this.pendingEntries, this.creationOptions);
+    this.archiveBuffer = built.archiveBuffer;
+    this.pendingEntries = undefined;
+    return this;
+  }
+
+  // ============================================
+  // EXTRACTION METHODS
+  // ============================================
+
   /**
-   * gets the original archive stream
+   * Get the original archive stream
    */
-  public async getArchiveStream() {
+  public async toStream(): Promise<plugins.stream.Readable> {
+    if (this.archiveBuffer) {
+      return plugins.stream.Readable.from(this.archiveBuffer);
+    }
     if (this.sourceStream) {
       return this.sourceStream;
     }
@@ -78,161 +270,320 @@ export class SmartArchive {
         .url(this.sourceUrl)
         .get();
       const webStream = response.stream();
-      // @ts-ignore - Web stream to Node.js stream conversion
-      const urlStream = plugins.stream.Readable.fromWeb(webStream);
-      return urlStream;
+      return plugins.stream.Readable.fromWeb(webStream as any);
     }
     if (this.sourceFilePath) {
-      const fileStream = plugins.fs.createReadStream(this.sourceFilePath);
-      return fileStream;
+      return plugins.fs.createReadStream(this.sourceFilePath);
+    }
+    throw new Error('No archive source configured');
+  }
+
+  /**
+   * Get archive as a Buffer
+   */
+  public async toBuffer(): Promise<Buffer> {
+    if (this.archiveBuffer) {
+      return this.archiveBuffer;
     }
+    const stream = await this.toStream();
+    return SmartArchive.streamToBuffer(stream);
   }
 
-  public async exportToTarGzStream() {
-    const tarPackStream = await this.tarTools.getPackStream();
-    const gzipStream = await this.gzipTools.getCompressionStream();
-    // const archiveStream = tarPackStream.pipe(gzipStream);
-    // return archiveStream;
+  /**
+   * Write archive to a file
+   */
+  public async toFile(filePath: string): Promise<void> {
+    const buffer = await this.toBuffer();
+    await plugins.fsPromises.mkdir(plugins.path.dirname(filePath), { recursive: true });
+    await plugins.fsPromises.writeFile(filePath, buffer);
   }
 
-  public async exportToFs(
+  /**
+   * Extract archive to filesystem
+   */
+  public async extractToDirectory(
     targetDir: string,
-    fileNameArg?: string,
+    options?: Partial<IArchiveExtractionOptions>
   ): Promise<void> {
     const done = plugins.smartpromise.defer<void>();
-    const streamFileStream = await this.exportToStreamOfStreamFiles();
+    const streamFileStream = await this.extractToStream();
+
     streamFileStream.pipe(
       new plugins.smartstream.SmartDuplex({
         objectMode: true,
-        writeFunction: async (
-          streamFileArg: plugins.smartfile.StreamFile,
-          streamtools,
-        ) => {
-          const done = plugins.smartpromise.defer<void>();
-          console.log(
-            streamFileArg.relativeFilePath
-              ? streamFileArg.relativeFilePath
-              : 'no relative path',
-          );
+        writeFunction: async (streamFileArg: plugins.smartfile.StreamFile) => {
+          const innerDone = plugins.smartpromise.defer<void>();
           const streamFile = streamFileArg;
+          let relativePath = streamFile.relativeFilePath || options?.fileName || 'extracted_file';
+
+          // Apply stripComponents if specified
+          if (options?.stripComponents && options.stripComponents > 0) {
+            const parts = relativePath.split('/');
+            relativePath = parts.slice(options.stripComponents).join('/');
+            if (!relativePath) {
+              innerDone.resolve();
+              return;
+            }
+          }
+
+          // Apply filter if specified
+          if (options?.filter) {
+            const entryInfo: IArchiveEntryInfo = {
+              path: relativePath,
+              size: 0,
+              isDirectory: false,
+              isFile: true,
+            };
+            if (!options.filter(entryInfo)) {
+              innerDone.resolve();
+              return;
+            }
+          }
+
           const readStream = await streamFile.createReadStream();
           await plugins.fsPromises.mkdir(targetDir, { recursive: true });
-          const writePath = plugins.path.join(
-            targetDir,
-            streamFile.relativeFilePath || fileNameArg,
-          );
+          const writePath = plugins.path.join(targetDir, relativePath);
           await plugins.fsPromises.mkdir(plugins.path.dirname(writePath), { recursive: true });
           const writeStream = plugins.fs.createWriteStream(writePath);
           readStream.pipe(writeStream);
           writeStream.on('finish', () => {
-            done.resolve();
+            innerDone.resolve();
           });
-          await done.promise;
+          await innerDone.promise;
         },
         finalFunction: async () => {
           done.resolve();
         },
-      }),
+      })
     );
+
     return done.promise;
   }
 
-  public async exportToStreamOfStreamFiles() {
-    const streamFileIntake =
-      new plugins.smartstream.StreamIntake<plugins.smartfile.StreamFile>({
-        objectMode: true,
-      });
-    const archiveStream = await this.getArchiveStream();
+  /**
+   * Extract archive to a stream of StreamFile objects
+   */
+  public async extractToStream(): Promise<plugins.smartstream.StreamIntake<plugins.smartfile.StreamFile>> {
+    const streamFileIntake = new plugins.smartstream.StreamIntake<plugins.smartfile.StreamFile>({
+      objectMode: true,
+    });
+
+    // Guard to prevent multiple signalEnd calls
+    let hasSignaledEnd = false;
+    const safeSignalEnd = () => {
+      if (!hasSignaledEnd) {
+        hasSignaledEnd = true;
+        streamFileIntake.signalEnd();
+      }
+    };
+
+    const archiveStream = await this.toStream();
     const createAnalyzedStream = () => this.archiveAnalyzer.getAnalyzedStream();
 
-    // lets create a function that can be called multiple times to unpack layers of archives
     const createUnpackStream = () =>
-      plugins.smartstream.createTransformFunction<IAnalyzedResult, any>(
+      plugins.smartstream.createTransformFunction<IAnalyzedResult, void>(
         async (analyzedResultChunk) => {
           if (analyzedResultChunk.fileType?.mime === 'application/x-tar') {
-            const tarStream =
-              analyzedResultChunk.decompressionStream as plugins.tarStream.Extract;
+            const tarStream = analyzedResultChunk.decompressionStream as plugins.tarStream.Extract;
+
             tarStream.on('entry', async (header, stream, next) => {
               if (header.type === 'directory') {
-                console.log(
-                  `tar stream directory: ${header.name} ... skipping!`,
-                );
-                stream.resume(); // Consume directory stream
+                stream.resume();
                 stream.on('end', () => next());
                 return;
               }
-              console.log(`tar stream file: ${header.name}`);
-              
-              // Create a PassThrough stream to buffer the data
+
               const passThrough = new plugins.stream.PassThrough();
-              const streamfile = plugins.smartfile.StreamFile.fromStream(
-                passThrough,
-                header.name,
-              );
-              
-              // Push the streamfile immediately
+              const streamfile = plugins.smartfile.StreamFile.fromStream(passThrough, header.name);
               streamFileIntake.push(streamfile);
-              
-              // Pipe the tar entry stream to the passthrough
               stream.pipe(passThrough);
-              
-              // Move to next entry when this one ends
               stream.on('end', () => {
                 passThrough.end();
                 next();
               });
             });
-            tarStream.on('finish', function () {
-              console.log('tar extraction finished');
-              // Only signal end if this is the final stream
-              streamFileIntake.signalEnd();
+
+            tarStream.on('finish', () => {
+              safeSignalEnd();
             });
-            analyzedResultChunk.resultStream.pipe(
-              analyzedResultChunk.decompressionStream,
-            );
+
+            analyzedResultChunk.resultStream.pipe(analyzedResultChunk.decompressionStream);
           } else if (analyzedResultChunk.fileType?.mime === 'application/zip') {
             analyzedResultChunk.resultStream
               .pipe(analyzedResultChunk.decompressionStream)
               .pipe(
                 new plugins.smartstream.SmartDuplex({
                   objectMode: true,
-                  writeFunction: async (
-                    streamFileArg: plugins.smartfile.StreamFile,
-                    streamtools,
-                  ) => {
+                  writeFunction: async (streamFileArg: plugins.smartfile.StreamFile) => {
                     streamFileIntake.push(streamFileArg);
                   },
                   finalFunction: async () => {
-                    streamFileIntake.signalEnd();
+                    safeSignalEnd();
                   },
-                }),
+                })
               );
-          } else if (
-            analyzedResultChunk.isArchive &&
-            analyzedResultChunk.decompressionStream
-          ) {
+          } else if (analyzedResultChunk.isArchive && analyzedResultChunk.decompressionStream) {
             // For nested archives (like gzip containing tar)
-            const nestedStream = analyzedResultChunk.resultStream
+            analyzedResultChunk.resultStream
               .pipe(analyzedResultChunk.decompressionStream)
               .pipe(createAnalyzedStream())
               .pipe(createUnpackStream());
-            
-            // Don't signal end here - let the nested unpacker handle it
           } else {
             const streamFile = plugins.smartfile.StreamFile.fromStream(
               analyzedResultChunk.resultStream,
-              analyzedResultChunk.fileType?.ext,
+              analyzedResultChunk.fileType?.ext
             );
             streamFileIntake.push(streamFile);
-            streamFileIntake.signalEnd();
+            safeSignalEnd();
           }
         },
-        {
-          objectMode: true,
-        },
+        { objectMode: true }
       );
 
     archiveStream.pipe(createAnalyzedStream()).pipe(createUnpackStream());
     return streamFileIntake;
   }
+
+  /**
+   * Extract archive to an array of SmartFile objects (in-memory)
+   */
+  public async extractToSmartFiles(): Promise<plugins.smartfile.SmartFile[]> {
+    const streamFiles = await this.extractToStream();
+    const smartFiles: plugins.smartfile.SmartFile[] = [];
+
+    return new Promise((resolve, reject) => {
+      streamFiles.on('data', async (streamFile: plugins.smartfile.StreamFile) => {
+        try {
+          const smartFile = await streamFile.toSmartFile();
+          smartFiles.push(smartFile);
+        } catch (err) {
+          reject(err);
+        }
+      });
+      streamFiles.on('end', () => resolve(smartFiles));
+      streamFiles.on('error', reject);
+    });
+  }
+
+  /**
+   * Extract a single file from the archive by path
+   */
+  public async extractFile(filePath: string): Promise<plugins.smartfile.SmartFile | null> {
+    const streamFiles = await this.extractToStream();
+
+    return new Promise((resolve, reject) => {
+      let found = false;
+
+      streamFiles.on('data', async (streamFile: plugins.smartfile.StreamFile) => {
+        if (streamFile.relativeFilePath === filePath || streamFile.relativeFilePath?.endsWith(filePath)) {
+          found = true;
+          try {
+            const smartFile = await streamFile.toSmartFile();
+            resolve(smartFile);
+          } catch (err) {
+            reject(err);
+          }
+        }
+      });
+
+      streamFiles.on('end', () => {
+        if (!found) {
+          resolve(null);
+        }
+      });
+
+      streamFiles.on('error', reject);
+    });
+  }
+
+  // ============================================
+  // ANALYSIS METHODS
+  // ============================================
+
+  /**
+   * Analyze the archive and return metadata
+   */
+  public async analyze(): Promise<IArchiveInfo> {
+    const stream = await this.toStream();
+    const firstChunk = await this.readFirstChunk(stream);
+    const fileType = await plugins.fileType.fileTypeFromBuffer(firstChunk);
+
+    let format: TArchiveFormat | null = null;
+    let isCompressed = false;
+    let isArchive = false;
+
+    if (fileType) {
+      switch (fileType.mime) {
+        case 'application/gzip':
+          format = 'gz';
+          isCompressed = true;
+          isArchive = true;
+          break;
+        case 'application/zip':
+          format = 'zip';
+          isCompressed = true;
+          isArchive = true;
+          break;
+        case 'application/x-tar':
+          format = 'tar';
+          isArchive = true;
+          break;
+        case 'application/x-bzip2':
+          format = 'bz2';
+          isCompressed = true;
+          isArchive = true;
+          break;
+      }
+    }
+
+    return {
+      format,
+      isCompressed,
+      isArchive,
+    };
+  }
+
+  /**
+   * List all entries in the archive without extracting
+   */
+  public async listEntries(): Promise<IArchiveEntryInfo[]> {
+    const entries: IArchiveEntryInfo[] = [];
+    const streamFiles = await this.extractToStream();
+
+    return new Promise((resolve, reject) => {
+      streamFiles.on('data', (streamFile: plugins.smartfile.StreamFile) => {
+        entries.push({
+          path: streamFile.relativeFilePath || 'unknown',
+          size: 0, // Size not available without reading
+          isDirectory: false,
+          isFile: true,
+        });
+      });
+      streamFiles.on('end', () => resolve(entries));
+      streamFiles.on('error', reject);
+    });
+  }
+
+  /**
+   * Check if a specific file exists in the archive
+   */
+  public async hasFile(filePath: string): Promise<boolean> {
+    const entries = await this.listEntries();
+    return entries.some((e) => e.path === filePath || e.path.endsWith(filePath));
+  }
+
+  /**
+   * Helper to read first chunk from stream
+   */
+  private async readFirstChunk(stream: plugins.stream.Readable): Promise<Buffer> {
+    return new Promise((resolve, reject) => {
+      const onData = (chunk: Buffer) => {
+        stream.removeListener('data', onData);
+        stream.removeListener('error', reject);
+        resolve(chunk);
+      };
+      stream.on('data', onData);
+      stream.on('error', reject);
+    });
+  }
 }