Merge pull request #1065 from serpilliere/fix_bytes_dll_names

Fix bytes dllname

13 files changed, 70 insertions, 67 deletions

diff --git a/example/symbol_exec/dse_crackme.py b/example/symbol_exec/dse_crackme.py
index 90774dc3..be9f4490 100644
--- a/example/symbol_exec/dse_crackme.py
+++ b/example/symbol_exec/dse_crackme.py
@@ -280,7 +280,7 @@ while todo:
         sb.run()
     except FinishOn as finish_info:
         print(finish_info.string)
-        if finish_info.string == b"OK":
+        if finish_info.string == "OK":
             # Stop if the expected result is found
             found = True
             break
diff --git a/miasm/analysis/sandbox.py b/miasm/analysis/sandbox.py
index e5595071..b8aaf788 100644
--- a/miasm/analysis/sandbox.py
+++ b/miasm/analysis/sandbox.py
@@ -51,6 +51,8 @@ class Sandbox(object):
         """
 
         # Initialize
+        if not isinstance(fname, bytes):
+            fname = fname.encode('utf8')
         self.fname = fname
         self.options = options
         if custom_methods is None:
@@ -185,7 +187,7 @@ class OS_Win(OS):
                    "ole32.dll", "urlmon.dll",
                    "ws2_32.dll", 'advapi32.dll', "psapi.dll",
                    ]
-    modules_path = "win_dll"
+    modules_path = b"win_dll"
 
     def __init__(self, custom_methods, *args, **kwargs):
         from miasm.jitter.loader.pe import vm_load_pe, vm_load_pe_libs,\
diff --git a/miasm/jitter/jitload.py b/miasm/jitter/jitload.py
index 0d8ab722..9fcb0b0a 100644
--- a/miasm/jitter/jitload.py
+++ b/miasm/jitter/jitload.py
@@ -439,7 +439,8 @@ class Jitter(object):
                self.vm.get_mem(tmp, 1) != b"\x00"):
             tmp += 1
             l += 1
-        return self.vm.get_mem(addr, l)
+        value = self.vm.get_mem(addr, l)
+        return value.decode('utf8')
 
     def get_str_unic(self, addr, max_char=None):
         """Get unicode str from vm.
@@ -455,14 +456,14 @@ class Jitter(object):
         s = s.decode("utf-16le")
         return s
 
-    def set_str_ansi(self, addr, s):
+    def set_str_ansi(self, addr, string):
         """Set an ansi string in memory"""
-        s = s + b"\x00"
-        self.vm.set_mem(addr, s)
+        string = (string + "\x00").encode('utf8')
+        self.vm.set_mem(addr, string)
 
-    def set_str_unic(self, addr, s):
+    def set_str_unic(self, addr, string):
         """Set an unicode string in memory"""
-        s = b"\x00".join(list(s)) + b'\x00' * 3
+        s = (string + "\x00").encode('utf-16le')
         self.vm.set_mem(addr, s)
 
     @staticmethod
diff --git a/miasm/jitter/loader/pe.py b/miasm/jitter/loader/pe.py
index a82b79f6..c779f508 100644
--- a/miasm/jitter/loader/pe.py
+++ b/miasm/jitter/loader/pe.py
@@ -266,8 +266,12 @@ def vm_load_pe_libs(vm, libs_name, libs, lib_path_base, **kargs):
     Return a dictionary Filename -> PE instances
     Extra arguments are passed to vm_load_pe_lib
     """
-    return {fname: vm_load_pe_lib(vm, fname, libs, lib_path_base, **kargs)
-            for fname in libs_name}
+    out = {}
+    for fname in libs_name:
+        if not isinstance(fname, bytes):
+            fname = fname.encode('utf8')
+        out[fname] = vm_load_pe_lib(vm, fname, libs, lib_path_base, **kargs)
+    return out
 
 
 def vm_fix_imports_pe_libs(lib_imgs, libs, lib_path_base,
diff --git a/miasm/os_dep/common.py b/miasm/os_dep/common.py
index 87602b3c..0b4d7e11 100644
--- a/miasm/os_dep/common.py
+++ b/miasm/os_dep/common.py
@@ -130,16 +130,10 @@ def unix_to_sbpath(path):
 def get_fmt_args(fmt, cur_arg, get_str, get_arg_n):
     idx = 0
     fmt = get_str(fmt)
-    if isinstance(fmt, bytes):
-        chars_format = b'%cdfsuxX'
-        char_percent = b'%'
-        char_string = b's'
-        output = b""
-    else:
-        chars_format = u'%cdfsuxX'
-        char_percent = u'%'
-        char_string = u's'
-        output = u""
+    chars_format = '%cdfsuxX'
+    char_percent = '%'
+    char_string = 's'
+    output = ""
 
     while True:
         if idx == len(fmt):
diff --git a/miasm/os_dep/linux/syscall.py b/miasm/os_dep/linux/syscall.py
index 353d61cf..7fede9f1 100644
--- a/miasm/os_dep/linux/syscall.py
+++ b/miasm/os_dep/linux/syscall.py
@@ -528,7 +528,7 @@ def sys_x86_64_getdents(jitter, linux_env):
         d_reclen = 8 * 2 + 2 + 1 + len(name) + 1
         d_off = cur_len + d_reclen
         entry = struct.pack("QqH", d_ino, d_off, d_reclen) + \
-                name + b"\x00" + struct.pack("B", d_type)
+                name.encode("utf8") + b"\x00" + struct.pack("B", d_type)
         assert len(entry) == d_reclen
         return entry
 
diff --git a/miasm/os_dep/linux_stdlib.py b/miasm/os_dep/linux_stdlib.py
index b2836881..3fa5b02e 100644
--- a/miasm/os_dep/linux_stdlib.py
+++ b/miasm/os_dep/linux_stdlib.py
@@ -153,7 +153,7 @@ def xxx_snprintf(jitter):
     output = get_fmt_args(jitter, fmt, cur_arg)
     output = output[:size - 1]
     ret = len(output)
-    jitter.vm.set_mem(args.string, output + b'\x00')
+    jitter.vm.set_mem(args.string, (output + '\x00').encode('utf8'))
     return jitter.func_ret_systemv(ret_addr, ret)
 
 
@@ -162,7 +162,7 @@ def xxx_sprintf(jitter):
     cur_arg, fmt = 2, args.fmt
     output = get_fmt_args(jitter, fmt, cur_arg)
     ret = len(output)
-    jitter.vm.set_mem(args.string, output + b'\x00')
+    jitter.vm.set_mem(args.string, (output + '\x00').encode('utf8'))
     return jitter.func_ret_systemv(ret_addr, ret)
 
 
@@ -171,14 +171,14 @@ def xxx_printf(jitter):
     cur_arg, fmt = 1, args.fmt
     output = get_fmt_args(jitter, fmt, cur_arg)
     ret = len(output)
-    stdout.write(output)
+    stdout.write(output.encode('utf8'))
     return jitter.func_ret_systemv(ret_addr, ret)
 
 
 def xxx_strcpy(jitter):
     ret_ad, args = jitter.func_args_systemv(["dst", "src"])
-    str_src = jitter.get_str_ansi(args.src) + b'\x00'
-    jitter.vm.set_mem(args.dst, str_src)
+    str_src = jitter.get_str_ansi(args.src) + '\x00'
+    jitter.vm.set_mem(args.dst, str_src.encode('utf8'))
     jitter.func_ret_systemv(ret_ad, args.dst)
 
 
diff --git a/miasm/os_dep/win_api_x86_32.py b/miasm/os_dep/win_api_x86_32.py
index ebf40cb0..5ef1b845 100644
--- a/miasm/os_dep/win_api_x86_32.py
+++ b/miasm/os_dep/win_api_x86_32.py
@@ -774,7 +774,7 @@ def kernel32_VirtualAlloc(jitter):
     if args.lpvoid == 0:
         alloc_addr = winobjs.heap.next_addr(args.dwsize)
         jitter.vm.add_memory_page(
-            alloc_addr, ACCESS_DICT[args.flprotect], "\x00" * args.dwsize,
+            alloc_addr, ACCESS_DICT[args.flprotect], b"\x00" * args.dwsize,
             "Alloc in %s ret 0x%X" % (whoami(), ret_ad))
     else:
         all_mem = jitter.vm.get_all_memory()
@@ -785,7 +785,7 @@ def kernel32_VirtualAlloc(jitter):
             alloc_addr = winobjs.heap.next_addr(args.dwsize)
             # alloc_addr = args.lpvoid
             jitter.vm.add_memory_page(
-                alloc_addr, ACCESS_DICT[args.flprotect], "\x00" * args.dwsize,
+                alloc_addr, ACCESS_DICT[args.flprotect], b"\x00" * args.dwsize,
                 "Alloc in %s ret 0x%X" % (whoami(), ret_ad))
 
     log.info('VirtualAlloc addr: 0x%x', alloc_addr)
@@ -2070,7 +2070,7 @@ def msvcrt_sprintf(jitter):
     ret_ad, args, output = msvcrt_sprintf_str(jitter, jitter.get_str_ansi)
     ret = len(output)
     log.info("sprintf() = '%s'" % (output))
-    jitter.vm.set_mem(args.string, output + b'\x00')
+    jitter.vm.set_mem(args.string, (output + '\x00').encode('utf8'))
     return jitter.func_ret_cdecl(ret_ad, ret)
 
 def msvcrt_swprintf(jitter):
diff --git a/miasm/os_dep/win_api_x86_32_seh.py b/miasm/os_dep/win_api_x86_32_seh.py
index 40f15480..d1be9ad2 100644
--- a/miasm/os_dep/win_api_x86_32_seh.py
+++ b/miasm/os_dep/win_api_x86_32_seh.py
@@ -253,9 +253,9 @@ def create_modules_chain(jitter, name2module):
                         fname)
             continue
         addr = base_addr + i * 0x1000
-        bpath = fname.replace('/', '\\')
+        bpath = fname.replace(b'/', b'\\')
         bname_str = os.path.split(fname)[1].lower()
-        bname_unicode = bname_str.encode("utf-16le")
+        bname_unicode = bname_str.decode('utf8').encode("utf-16le")
         log.info("Add module %x %r", pe_obj.NThdr.ImageBase, bname_str)
 
         modules_info.add(bname_str, pe_obj, addr)
@@ -287,6 +287,8 @@ def create_modules_chain(jitter, name2module):
             "Module name %r" % bname_str
         )
 
+        if isinstance(bpath, bytes):
+            bpath = bpath.decode('utf8')
         bpath_unicode = bpath.encode('utf-16le')
         jitter.vm.add_memory_page(
             addr + offset_path,
diff --git a/test/arch/x86/qemu/testqemu.py b/test/arch/x86/qemu/testqemu.py
index 99d6e6c1..594a826b 100644
--- a/test/arch/x86/qemu/testqemu.py
+++ b/test/arch/x86/qemu/testqemu.py
@@ -16,24 +16,24 @@ from miasm.jitter.csts import PAGE_READ, PAGE_WRITE
 
 # Utils
 def parse_fmt(s):
-    fmt = s[:]+b"\x00"
+    fmt = s[:]+"\x00"
     out = []
     i = 0
     while i < len(fmt):
         c = fmt[i:i+1]
-        if c != b"%":
+        if c != "%":
             i+=1
             continue
-        if fmt[i+1:i+2] == b"%":
+        if fmt[i+1:i+2] == "%":
             i+=2
             continue
         j = 0
         i+=1
-        while fmt[i+j:i+j+1] in b"0123456789$.-":
+        while fmt[i+j:i+j+1] in "0123456789$.-":
             j+=1
-        if fmt[i+j:i+j+1] in [b'l']:
+        if fmt[i+j:i+j+1] in ['l']:
             j +=1
-        if fmt[i+j:i+j+1] == b"h":
+        if fmt[i+j:i+j+1] == "h":
             x = fmt[i+j:i+j+2]
         else:
             x = fmt[i+j:i+j+1]
@@ -50,8 +50,8 @@ def xxx___printf_chk(jitter):
         raise RuntimeError("Not implemented")
     fmt = jitter.get_str_ansi(args.format)
     # Manage llx
-    fmt = fmt.replace(b"llx", b"lx")
-    fmt = fmt.replace(b"%016lx", b"%016z")
+    fmt = fmt.replace("llx", "lx")
+    fmt = fmt.replace("%016lx", "%016z")
 
     fmt_a = parse_fmt(fmt)
     esp = jitter.cpu.ESP
@@ -59,15 +59,15 @@ def xxx___printf_chk(jitter):
     i = 0
     for x in fmt_a:
         a = jitter.vm.get_u32(esp + 8 + 4*i)
-        if x == b"s":
+        if x == "s":
             a = jitter.get_str_ansi(a)
-        elif x in (b"x", b'X', b"d"):
+        elif x in ("x", 'X', "d"):
             pass
-        elif x.lower() in (b"f", b"l"):
+        elif x.lower() in ("f", "l"):
             a2 = jitter.vm.get_u32(esp + 8 + 4*(i+1))
             a = struct.unpack("d", struct.pack("Q", a2 << 32 | a))[0]
             i += 1
-        elif x.lower() == b'z':
+        elif x.lower() == 'z':
             a2 = jitter.vm.get_u32(esp + 8 + 4*(i+1))
             a = a2 << 32 | a
             i += 1
@@ -75,22 +75,22 @@ def xxx___printf_chk(jitter):
             raise RuntimeError("Not implemented format")
         args.append(a)
         i += 1
-    fmt = fmt.replace(b"%016z", b"%016lx")
+    fmt = fmt.replace("%016z", "%016lx")
     output = fmt%(tuple(args))
     # NaN bad repr in Python
-    output = output.replace(b"nan", b"-nan")
+    output = output.replace("nan", "-nan")
 
-    if b"\n" not in output:
+    if "\n" not in output:
         raise RuntimeError("Format must end with a \\n")
 
     # Check with expected result
     line = next(expected)
-    if output != line.encode():
+    if output != line:
         print("Expected:", line)
         print("Obtained:", output)
         raise RuntimeError("Bad semantic")
 
-    stdout.write(b"[%d] %s" % (nb_tests, output))
+    stdout.write(b"[%d] %s" % (nb_tests, output.encode('utf8')))
     nb_tests += 1
     jitter.func_ret_systemv(ret_ad, 0)
 
@@ -105,7 +105,7 @@ def xxx_puts(jitter):
     output = jitter.get_str_ansi(args.target)
     # Check with expected result
     line = next(expected)
-    if output != line.rstrip().encode():
+    if output != line.rstrip():
         print("Expected:", line)
         print("Obtained:", output)
         raise RuntimeError("Bad semantic")
diff --git a/test/arch/x86/qemu/testqemu64.py b/test/arch/x86/qemu/testqemu64.py
index 24193d40..636cb6a9 100644
--- a/test/arch/x86/qemu/testqemu64.py
+++ b/test/arch/x86/qemu/testqemu64.py
@@ -16,24 +16,24 @@ from miasm.jitter.csts import PAGE_READ, PAGE_WRITE
 
 # Utils
 def parse_fmt(s):
-    fmt = s[:]+b"\x00"
+    fmt = s[:]+"\x00"
     out = []
     i = 0
     while i < len(fmt):
         c = fmt[i:i+1]
-        if c != b"%":
+        if c != "%":
             i+=1
             continue
-        if fmt[i+1:i+2] == b"%":
+        if fmt[i+1:i+2] == "%":
             i+=2
             continue
         j = 0
         i+=1
-        while fmt[i+j:i+j+1] in b"0123456789$.-":
+        while fmt[i+j:i+j+1] in "0123456789$.-":
             j+=1
-        if fmt[i+j:i+j+1] in [b'l']:
+        if fmt[i+j:i+j+1] in ['l']:
             j +=1
-        if fmt[i+j:i+j+1] == b"h":
+        if fmt[i+j:i+j+1] == "h":
             x = fmt[i+j:i+j+2]
         else:
             x = fmt[i+j:i+j+1]
@@ -50,8 +50,8 @@ def xxx___printf_chk(jitter):
         raise RuntimeError("Not implemented")
     fmt = jitter.get_str_ansi(args.format)
     # Manage llx
-    fmt = fmt.replace(b"llx", b"lx")
-    fmt = fmt.replace(b"%016lx", b"%016z")
+    fmt = fmt.replace("llx", "lx")
+    fmt = fmt.replace("%016lx", "%016z")
 
     fmt_a = parse_fmt(fmt)
     args = []
@@ -59,11 +59,11 @@ def xxx___printf_chk(jitter):
 
     for x in fmt_a:
         a = jitter.get_arg_n_systemv(2 + i)
-        if x == b"s":
+        if x == "s":
             a = jitter.get_str_ansi(a)
-        elif x in (b"x", b'X', b'd', b'z', b'Z'):
+        elif x in ("x", 'X', 'd', 'z', 'Z'):
             pass
-        elif x.lower() in (b"f","l"):
+        elif x.lower() in ("f","l"):
             a = struct.unpack("d", struct.pack("Q", a))[0]
             i += 1
         else:
@@ -71,22 +71,22 @@ def xxx___printf_chk(jitter):
         args.append(a)
         i += 1
 
-    fmt = fmt.replace(b"%016z", b"%016lx")
+    fmt = fmt.replace("%016z", "%016lx")
     output = fmt%(tuple(args))
     # NaN bad repr in Python
-    output = output.replace(b"nan", b"-nan")
+    output = output.replace("nan", "-nan")
 
-    if b"\n" not in output:
+    if "\n" not in output:
         raise RuntimeError("Format must end with a \\n")
 
     # Check with expected result
     line = next(expected)
-    if output != line.encode():
+    if output != line:
         print("Expected:", line)
         print("Obtained:", output)
         raise RuntimeError("Bad semantic")
 
-    stdout.write(b"[%d] %s" % (nb_tests, output))
+    stdout.write(b"[%d] %s" % (nb_tests, output.encode('utf8')))
     nb_tests += 1
     jitter.func_ret_systemv(ret_ad, 0)
 
diff --git a/test/os_dep/linux/stdlib.py b/test/os_dep/linux/stdlib.py
index a205002b..ef890625 100755
--- a/test/os_dep/linux/stdlib.py
+++ b/test/os_dep/linux/stdlib.py
@@ -34,7 +34,7 @@ class TestLinuxStdlib(unittest.TestCase):
         jit.push_uint32_t(0) # ret_ad
         stdlib.xxx_sprintf(jit)
         ret = jit.get_str_ansi(buf)
-        self.assertEqual(ret, b"'coucou' 1111")
+        self.assertEqual(ret, "'coucou' 1111")
 
 
 if __name__ == '__main__':
diff --git a/test/os_dep/win_api_x86_32.py b/test/os_dep/win_api_x86_32.py
index a7d88f90..f759c6af 100755
--- a/test/os_dep/win_api_x86_32.py
+++ b/test/os_dep/win_api_x86_32.py
@@ -43,7 +43,7 @@ class TestWinAPI(unittest.TestCase):
         jit.push_uint32_t(0) # ret_ad
         winapi.msvcrt_sprintf(jit)
         ret = jit.get_str_ansi(buf)
-        self.assertEqual(ret, b"'coucou' 1111")
+        self.assertEqual(ret, "'coucou' 1111")
 
 
     def test_msvcrt_swprintf(self):