Add metadata!

src/compress.cpp

@@ -0,0 +1,83 @@
+#include <fstream>
+#include <vector>
+#include <filesystem>
+#include <string>
+#include <random>
+#include <chrono>
+#include <stdexcept>
+#include <iostream> // For error reporting in TempDirectory destructor
+
+#include "compress.hpp"
+#include "hash.hpp"
+#include "temp_directory.hpp"
+
+
+namespace simple_object_storage {
+
+std::string decompress_gzip(const std::string& file_path) {
+    std::ifstream file(file_path, std::ios::binary);
+    if (!file) return {};
+
+    std::vector<char> compressed((std::istreambuf_iterator<char>(file)), std::istreambuf_iterator<char>());
+    if (compressed.size() < 2) return {};
+
+    // Skip gzip header (10 bytes)
+    size_t pos = 10;
+    if (compressed.size() <= pos) return {};
+
+    // Prepare zlib stream
+    z_stream strm = {};
+    strm.next_in = reinterpret_cast<Bytef*>(compressed.data() + pos);
+    strm.avail_in = compressed.size() - pos;
+
+    if (inflateInit2(&strm, 16 + MAX_WBITS) != Z_OK) return {};
+
+    std::string out;
+    char buffer[4096];
+    int ret;
+    do {
+        strm.next_out = reinterpret_cast<Bytef*>(buffer);
+        strm.avail_out = sizeof(buffer);
+        ret = inflate(&strm, Z_NO_FLUSH);
+        if (ret == Z_STREAM_ERROR || ret == Z_DATA_ERROR || ret == Z_MEM_ERROR) {
+            inflateEnd(&strm);
+            return {};
+        }
+        out.append(buffer, sizeof(buffer) - strm.avail_out);
+    } while (ret != Z_STREAM_END);
+
+    inflateEnd(&strm);
+    return out;
+}
+
+// if the file is a tgz file (we can't rely on the extension), then unpack on disk and has the contents
+// with hash_directory_recursive in hash.hpp
+uint64_t get_hash_from_tgz(const std::string &file_path)
+{
+    // check if it's a gzip file
+    std::ifstream file(file_path, std::ios::binary);
+    if (!file) return 0;
+
+    char buffer[2];
+    file.read(buffer, 2);
+    if (buffer[0] != 0x1F || buffer[1] != 0x8B) return 0;
+
+    // gunzip the file to a new temporary directory
+    TempDirectory temp_dir_manager("tgz_unpack_"); // Creates dir and schedules cleanup
+    std::string temp_dir = temp_dir_manager.string(); // Get the path string to use
+
+    std::string decompressed = decompress_gzip(file_path);
+
+    // unpack the file on disk
+    std::string command = "tar -xzf " + file_path + " -C " + temp_dir;
+    int result = system(command.c_str()); // Basic tar extraction - requires 'tar' command
+    if (result != 0) {
+        std::cerr << "Error unpacking tgz file: " << file_path << std::endl;
+        return 0;
+    }
+
+    // hash the contents
+    return hash_directory_recursive(temp_dir);
+}
+
+} // namespace simple_object_storage