NonLLVMMisc: Add FileHandling class

This class will contain a static method to detect which file we are
dealing with w.r.t ELF files.

Signed-off-by: Marcos Paulo de Souza <mpdesouza@suse.com>

libcextract/ElfCXX.cpp

@@ -91,12 +91,23 @@ ElfObject::ElfObject(const char *path)
     throw std::runtime_error("ELF file not found: " + parser_path);
   }
 
-  /* Create libelf object and store it in the class variable.  */
-  ElfObj = elf_begin(ElfFd, ELF_C_READ, nullptr);
-  if (ElfObj == nullptr) {
+  enum FileHandling::FileType ft = FileHandling::Get_File_Type(ElfFd);
+
+  switch (ft) {
+  case FileHandling::FILE_TYPE_ELF: {
+    /* Create libelf object and store it in the class variable.  */
+    ElfObj = elf_begin(ElfFd, ELF_C_READ, nullptr);
+    if (ElfObj == nullptr) {
+      close(ElfFd);
+      throw std::runtime_error("libelf error on file " + parser_path + ": "
+                                   + std::string(elf_errmsg(elf_errno())));
+    }
+    break;
+  }
+
+  default:
     close(ElfFd);
-    throw std::runtime_error("libelf error on file " + parser_path + ": "
-                                 + std::string(elf_errmsg(elf_errno())));
+    throw std::runtime_error("Format not recognized: " + parser_path + "\n");
   }
 }
 

libcextract/NonLLVMMisc.cpp

@@ -24,6 +24,9 @@
 #include <sys/stat.h>
 #include <unistd.h>
 #include <errno.h>
+#include <fcntl.h>
+
+#include <iostream>
 
 /** @brief Handle some quirks of getline.  */
 char *getline_easy(FILE *file)
@@ -102,3 +105,35 @@ bool check_color_available(void)
   const char *term = getenv("TERM");
   return term && strcmp(term, "dumb") != 0 && isatty(STDOUT_FILENO);
 }
+
+enum FileHandling::FileType FileHandling::Get_File_Type(int fd)
+{
+
+  unsigned char buf[4] = { 0 };
+
+  int ret = read(fd, buf, 4);
+  if (ret < 0) {
+   throw std::runtime_error("error reading ELF file: " + std::to_string(ret));
+  } else if (ret == 0) {
+   throw std::runtime_error("error reading ELF file: empty file?");
+  }
+
+  /* reset file pointer */
+  lseek(fd, 0, SEEK_SET);
+
+  const unsigned char elf_magic[4] =  { 0x7f, 0x45, 0x4c, 0x46 };
+  const unsigned char zlib_magic[2] = { 0x1f, 0x8b };
+  const unsigned char zstd_magic[4] = { 0x28, 0xb5, 0x2f, 0xfd };
+
+  /* Check if the file is an ELF */
+  if (!memcmp(buf, elf_magic, 4))
+    return FILE_TYPE_ELF;
+
+  else if (!memcmp(buf, zlib_magic, 2))
+    return FILE_TYPE_GZ;
+
+  else if (!memcmp(buf, zstd_magic, 4))
+    return FILE_TYPE_ZSTD;
+
+  return FILE_TYPE_UNKNOWN;
+}

libcextract/NonLLVMMisc.hh

@@ -115,3 +115,16 @@ class VectorRef
   T *Ref;
   unsigned Size;
 };
+
+class FileHandling
+{
+  public:
+  enum FileType {
+    FILE_TYPE_ELF,
+    FILE_TYPE_GZ,
+    FILE_TYPE_ZSTD,
+    FILE_TYPE_UNKNOWN,
+  };
+
+  static enum FileType Get_File_Type(int fd);
+};