Installation

pip install objutils

For development (editable install):

pip install -e .

Or with Poetry:

poetry install

Prerequisites

• Python >= 3.9

Features

• Read ELF files (including symbols) and extract loadable sections.

• Inspect PE/COFF files and symbols (optional PDB support).

• Merge multiple HEX files into one (oj-hex-merge).

• Split HEX files into separate sections (oj-hex-split).

• Typed access (scalars, arrays, strings) to binary data.

Recent improvements

• ElfParser now supports in_memory=True: use a transient :memory: SQLite database instead of persisting a .prgdb file – ideal for read-only analysis, CI pipelines and unit tests.

• New oj-hex-merge and oj-hex-split CLI tools for combining and partitioning HEX files.

• New oj-dwarf-import CLI replaces dwarfer.py/cu_info.py for DWARF imports, CU listing, summaries, and attribute traversal.

• Faster DWARF imports: batched ORM writes, quiet flag propagation, and safe DWARF expression evaluation keep large ELF files stable.

• Exception handling and typing tightened across DWARF/ELF/PECOFF modules and public APIs to surface real errors without masking them.

• ElfParser.close() releases SQLite/mmap handles; examples/tests now close parsers to avoid database locks.

• S-Record metadata stays minimal and numeric (no data records mixed in); Mostec/Tek SREC roundtrips now reproduce expected files.

• Fortran-ordered ndarray reads/writes honor legacy column-major expectations; hexdump empty rows keep canonical spacing.

img = objutils.load("ihex", "myHexFile.hex")     # Load an Intel HEX file...
objutils.dump("srec", "mySRecFile.srec", img)    # and save it as S-Records.

First steps

If you are interested, what objutils provides to you out-of-the-box, refer to Scripts documentation.

In any case, you should work through the following tutorial:

First import all classes and functions used in this tutorial.

from objutils import Image, Section, dump, dumps, load, loads

Everything starts with hello world…

sec0 = Section(start_address = 0x1000, data = "Hello HEX world!")

The constructor parameters to Section reflect what they are about: A continuous area of memory with an start address.

data is not necessarily a string, array.array**s, **byte, bytearray will also do, or from an internal point of view: everything that is convertible to bytearray could be used.

Note: start_address and data are positional arguments, so there is no need to use them as keywords (just for the sake of illustration).

Now let’s inspect our section.

sec0.hexdump()

00001000  48 65 6c 6c 6f 20 48 45 58 20 77 6f 72 6c 64 21  |Hello HEX world!|
---------------
       16 bytes
---------------

hexdump() gives us, what in the world of hackers is known as a canonical hexdump.

HEX files usually consist of more than one section, so let’s create another one.

sec1 = Section(0x2000, range(1, 17))
sec1.hexdump()

00002000  01 02 03 04 05 06 07 08 09 0a 0b 0c 0d 0e 0f 10  |................|
---------------
       16 bytes
---------------

Now, let’s glue together our sections.

img0 = Image([sec0, sec1])
print(img0)

Section(address = 0X00001000, length = 16, data = b'Hello HEX world!')
Section(address = 0X00002000, length = 16, data = b'\x01\x02\x03\x04\x05\x06\x07\x08\t\n\x0b\x0c\r\x0e\x0f\x10')

Images are obviously a container for sections, and they are always involved if you are interacting with disk based HEX files.

dump("srec", "example0.srec", img0)

The resulting file could be inspected from command line.

$ cat example0.srec
S113100048656C6C6F2048455820776F726C64217A
S11320000102030405060708090A0B0C0D0E0F1044

And loaded again…

img1 = load("srec", "example0.srec")
print(img1)

Section(address = 0X00001000, length = 16, data = b'Hello HEX world!')
Section(address = 0X00002000, length = 16, data = b'\x01\x02\x03\x04\x05\x06\x07\x08\t\n\x0b\x0c\r\x0e\x0f\x10')

Or split it into its sections…

images = img1.split()
for i, img in enumerate(images):
    dump("srec", f"part_{i}.srec", img)

This leads to the conversion idiom.

img1 = load("srec", "example0.srec")
dump("ihex", "example0.hex", img1)

Note: the formats above listed as historical are for one good reason historical: they are only 16bit wide, so if you want to convert, say a srec file for a 32bit MCU to them, you’re out of luck.

OK, we’re starting another session.

sec0 = Section(0x100, range(1, 9))
sec1 = Section(0x108, range(9, 17))
img0 = Image([sec0, sec1])
print(img0)

Section(address = 0X00000100, length = 16, data = b'\x01\x02\x03\x04\x05\x06\x07\x08\t\n\x0b\x0c\r\x0e\x0f\x10')

img0.hexdump()

Section #0000
-------------
00000100  01 02 03 04 05 06 07 08 09 0a 0b 0c 0d 0e 0f 10  |................|
---------------
       16 bytes
---------------

Two sections with consecutive address ranges concatenated to one, this may or may not what you are expected.

For this reason Image has a join parameter.

sec0 = Section(0x100, range(1, 9))
sec1 = Section(0x108, range(9, 17))
img0 = Image([sec0, sec1], join = False)
print(img0)

Section(address = 0X00000100, length = 8, data = b'\x01\x02\x03\x04\x05\x06\x07\x08')
Section(address = 0X00000108, length = 8, data = b'\t\n\x0b\x0c\r\x0e\x0f\x10')

img0.hexdump()

Section #0000
-------------
00000100  01 02 03 04 05 06 07 08                          |........        |
---------------
        8 bytes
---------------

Section #0001
-------------
00000108  09 0a 0b 0c 0d 0e 0f 10                          |........        |
---------------
        8 bytes
---------------

One feature that sets objutils apart from other libraries of this breed is typified access.

We are starting with a new image.

img0 = Image([Section(0x1000, bytes(64))])
print(img0)

Section(address = 0X00001000, length = 64, data = b'\x00\x00\x00\x00\x00\x00\x00...00\x00\x00\x00\x00\x00\x00\x00')

We are now writing a string to our image.

img0 = Image([Section(0x1000, bytes(64))])
img0.write(0x1010, [0xff])
img0.hexdump()

Section #0000
-------------
00001000  00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00  |................|
00001010  ff 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00  |................|
00001020  00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00  |................|
00001030  00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00  |................|

img0.write_string(0x1000, "Hello HEX world!")
img0.hexdump()

Section #0000
-------------
00001000  48 65 6c 6c 6f 20 48 45 58 20 77 6f 72 6c 64 21  |Hello HEX world!|
00001010  00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00  |................|
          *
00001030  00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00  |................|
---------------
       64 bytes
---------------

Notice the difference? In our Section example above, the string passed as a data parameter was just a bunch of bytes, but now it is a “real” C-string (there is a opposite function, read_string, that scans for a terminating NULL character).

Use write() and read() functions, if you want to access plain bytes.

But there is also support for numerical types.

img0 = Image([Section(0x1000, bytes(64))])
img0.write_numeric(0x1000, 0x10203040, "uint32_be")
img0.write_numeric(0x1004, 0x50607080, "uint32_le")
img0.hexdump()

Section #0000
-------------
00001000  10 20 30 40 80 70 60 50 00 00 00 00 00 00 00 00  |. 0@.p`P........|
00001010  00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00  |................|
          *
00001030  00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00  |................|
---------------
       64 bytes
---------------

The folling types are supported:

• uint8

• int8

• uint16

• int16

• uint32

• int32

• uint64

• int64

• float32

• float64

In any case, endianess suffixes _be or _le are required.

For ASAM workflows there are dedicated helpers with explicit byte-order names:

img0 = Image([Section(0x1000, bytes(64))])
img0.write_asam_numeric(0x1000, 0x11223344, "ULONG", "MSB_FIRST")
img0.write_asam_numeric(0x1004, 0x11223344, "ULONG", "MSB_FIRST_MSW_LAST")
img0.write_asam_numeric(0x1008, 0x11223344, "ULONG", "MSB_LAST_MSW_FIRST")

print(hex(img0.read_asam_numeric(0x1000, "ULONG", "MSB_FIRST")))
print(hex(img0.read_asam_numeric(0x1004, "ULONG", "MSB_FIRST_MSW_LAST")))
print(hex(img0.read_asam_numeric(0x1008, "ULONG", "MSB_LAST_MSW_FIRST")))

# All reads print 0x11223344 again.

Supported ASAM byte orders:

• MSB_FIRST

• MSB_LAST

• MSB_FIRST_MSW_LAST (word-swap)

• MSB_LAST_MSW_FIRST (word-swap)

• LITTLE_ENDIAN (legacy alias for MSB_LAST)

• BIG_ENDIAN (legacy alias for MSB_FIRST)

Supported ASAM numeric datatypes:

• UBYTE, SBYTE

• UWORD, SWORD

• ULONG, SLONG

• A_UINT64, A_INT64

• FLOAT16_IEEE, FLOAT32_IEEE, FLOAT64_IEEE

ASAM string helpers are available, too:

img0.write_asam_string(0x1020, "MOTOR", "ASCII")
img0.write_asam_string(0x1030, "Drehzahl", "UTF8")

print(img0.read_asam_string(0x1020, "ASCII"))
print(img0.read_asam_string(0x1030, "UTF8"))

Supported ASAM string datatypes:

• ASCII

• UTF8

• UTF16

• UTF32

Arrays are also supported.

img0 = Image([Section(0x1000, bytes(64))])
img0.write_numeric_array(0x1000, [0x1000, 0x2000, 0x3000, 0x4000, 0x5000, 0x6000, 0x7000, 0x8000], "uint16_le")
img0.hexdump()

Section #0000
-------------
00001000  00 10 00 20 00 30 00 40 00 50 00 60 00 70 00 80  |... .0.@.P.`.p..|
00001010  00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00  |................|
          *
00001030  00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00  |................|
---------------
       64 bytes
---------------

This concludes our tutorial for now, but there is more stuff to follow…

Documentation

For full documentation, including installation, tutorials and PDF documents, please see Readthedocs

Bugs/Requests

Please use the GitHub issue tracker to submit bugs or request features

References

Here is an overview of some of the classic hex-file formats.

Authors

• Christoph Schueler - Initial work and project lead.

License

This project is licensed under the GNU General Public License v2.0

Contribution

If you contribute code to this project, you are implicitly allowing your code to be distributed under the GNU General Public License v2.0. You are also implicitly verifying that all code is your original work.