Add bz2 decompression for non archive file (emscripten-forge#34)

* Add bz2 decompression for non archive file

* Refactoring

* Fix eslint

* Fix typos

Author: AnastasiaSliusar

src/index.ts

@@ -13,7 +13,10 @@ const fetchByteArray = async (url: string): Promise<Uint8Array> => {
 export const initUntarJS = async (): Promise<IUnpackJSAPI> => {
   const wasmModule = await initializeWasm();
 
-  const extractData = async (data: Uint8Array): Promise<FilesData> => {
+  const extractData = async (
+    data: Uint8Array,
+    decompressionOnly: boolean = false
+  ): Promise<FilesData> => {
     /**Since WebAssembly, memory is accessed using pointers
       and the first parameter of extract_archive method from unpack.c, which is Uint8Array of file data, should be a pointer
       so we have to allocate memory for file data
@@ -27,9 +30,10 @@ export const initUntarJS = async (): Promise<IUnpackJSAPI> => {
     let resultPtr: number | null = wasmModule._extract_archive(
       inputPtr,
       data.length,
-      fileCountPtr
+      fileCountPtr,
+      decompressionOnly
     );
-
+    const files: FilesData = {};
     /**
      * Since extract_archive returns a pointer that refers to an instance of the ExtractedArchive in unpack.c
         typedef struct {
@@ -71,8 +75,6 @@ export const initUntarJS = async (): Promise<IUnpackJSAPI> => {
     const filesPtr = wasmModule.getValue(resultPtr, 'i32');
     const fileCount = wasmModule.getValue(resultPtr + 4, 'i32');
 
-    const files: FilesData = {};
-
     /**
      * FilesPtr is a pointer that refers to an instance of the FileData in unpack.c
         typedef struct {
@@ -102,26 +104,38 @@ export const initUntarJS = async (): Promise<IUnpackJSAPI> => {
         dataPtr,
         dataSize
       );
-      
-      const fileDataCopy = fileData.slice(0);
 
+      const fileDataCopy = fileData.slice(0);
       files[filename] = fileDataCopy;
     }
-  
+
     wasmModule._free(inputPtr);
     wasmModule._free(fileCountPtr);
     wasmModule._free_extracted_archive(resultPtr);
     inputPtr = null;
     fileCountPtr = null;
     resultPtr = null;
     errorMessagePtr = null;
+
     return files;
   };
 
   const extract = async (url: string): Promise<FilesData> => {
+    let isArchive: boolean = checkIsArchive(url);
     const data = await fetchByteArray(url);
-    return extractData(data);
-  }
+    return extractData(data, !isArchive);
+  };
+
+  const checkIsArchive = (url: string): boolean => {
+    let isArchive: boolean = false;
+    let archiveExtArr = ['.conda', 'tar.bz2', 'tar.gz'];
+    archiveExtArr.forEach(type => {
+      if (url.toLowerCase().endsWith(type)) {
+        isArchive = true;
+      }
+    });
+    return isArchive;
+  };
 
   return {
     extract,

src/unpack.d.ts

@@ -8,7 +8,8 @@ export interface IWasmModule {
   _extract_archive(
     inputPtr: number,
     inputSize: number,
-    fileCountPtr: number
+    fileCountPtr: number,
+    decompressionOnly: boolean
   ): number;
   getValue(ptr: number, type: string): number;
 }

unpack.c

@@ -1,6 +1,7 @@
 #include <stdio.h>
 #include <string.h>
 #include <stdlib.h>
+#include <stdbool.h>
 #include <archive.h>
 #include <archive_entry.h>
 #include <emscripten.h>
@@ -18,8 +19,9 @@ typedef struct {
     char error_message[256];
 } ExtractedArchive;
 
+
 EMSCRIPTEN_KEEPALIVE
-ExtractedArchive* extract_archive(uint8_t* inputData, size_t inputSize, size_t* fileCount) {
+ExtractedArchive* extract_archive(uint8_t* inputData, size_t inputSize, size_t* fileCount, bool decompressionOnly ) {
     struct archive* archive;
     struct archive_entry* entry;
     FileData* files = NULL;
@@ -38,6 +40,9 @@ ExtractedArchive* extract_archive(uint8_t* inputData, size_t inputSize, size_t*
     archive = archive_read_new();
     archive_read_support_filter_all(archive);
     archive_read_support_format_all(archive);
+    if (decompressionOnly) {
+        archive_read_support_format_raw(archive);
+    }
 
     if (archive_read_open_memory(archive, inputData, inputSize) != ARCHIVE_OK) {
         result->status = 0;
@@ -47,10 +52,8 @@ ExtractedArchive* extract_archive(uint8_t* inputData, size_t inputSize, size_t*
     }
 
     while (archive_read_next_header(archive, &entry) == ARCHIVE_OK) {
-        const char* filename = archive_entry_pathname(entry);
-        size_t entrySize = archive_entry_size(entry);
-
-   
+        const char* filename =  decompressionOnly ? "decompression.json": archive_entry_pathname(entry);
+        size_t entrySize = decompressionOnly ? inputSize: archive_entry_size(entry);
         files= realloc(files, sizeof(FileData) * (files_count + 1));
         if (!files) {
             archive_read_free(archive);
@@ -96,7 +99,6 @@ ExtractedArchive* extract_archive(uint8_t* inputData, size_t inputSize, size_t*
     return result;
 }
 
-
 EMSCRIPTEN_KEEPALIVE
 void free_extracted_archive(ExtractedArchive* archive) {
     if (!archive) {