Extracts MStar/MediaTek Widevine L1 keyboxes and PlayReady passphrases from raw eMMC dumps and firmware images. Only supports MStar/MediaTek SoC platforms (LG webOS, Hisense, Sharp, Philips/TPV, Panasonic, etc).

Requirements

Python 3.8+

Install everything once:

pip install -r requirements.txt

This pulls in pycryptodome, requests, and pywidevine.

Per-script breakdown:

kbxtractor has no external dependencies
kbxdecoder needs pycryptodome for decryption, requests for URL key sources, and pywidevine for --extract-wvd
prxtractor needs pycryptodome

kbxtractor

Scans raw eMMC dumps and firmware images for MStar Widevine L1 keyboxes and extracts them as encrypted .bin files.

How?

Searches a binary for MSTAR_SECURE_STORE_FILE_MAGIC_ID prefix and extracts 228 bytes from matches and filters them with zero-byte density, maximum zero runs, and tail entropy.

Usage

Scan a single file:

python kbxtractor.py firmware.bin

Scan all files in the script's directory:

python kbxtractor.py --all

Save output to a specific folder:

python kbxtractor.py firmware.bin -o ./output

Output

Found keyboxes are saved as:

<source>_Keybox.bin
<source>_Keybox_2.bin

A hex dump of each hit is printed to the console by default.

All flags

Flag	Description
`files`	One or more binary files to scan
`--all`	Scan all files in the current directory
`-o`, `--output-dir`	Directory to save extracted keyboxes
`--no-hexview`	Suppress hex dump output
`--verbose-filters`	Print the reason for rejections
`-q`, `--quiet`	Suppress all output except errors

kbxdecoder

Scans raw eMMC dumps and firmware images for MStar Widevine L1 keyboxes, attempts AES decryption using supplied keys. Can generate .wvd files.

How?

Attempts to decrypt raw keybox blobs with ECB, CBC, CFB, OFB, and CTR AES modes. A successful decryption is recognised by the INNER_MSTAR marker. Finds the 128-byte Widevine keybox structure, extracts the Device ID, checks the CRC-32, then saves the keybox.

Key sourcing

The --key flag accepts:

A hex string - Single AES key (32, 48, or 64 hex characters for AES-128/192/256)
A file path - Local .txt file containing one key per line, or KEY,IV pairs for CBC
A URL - Remote key list

Pass --key multiple times to combine. URL-sourced keys are cached to keys.txt.

Usage

Scan with keys from a local file:

python kbxdecoder.py firmware.bin --key keys.txt

Scan with keys from a URL:

python kbxdecoder.py firmware.bin --key https://example.com/keys.txt

Scan with a single manual key:

python kbxdecoder.py firmware.bin --key <KEY>

Combine sources:

python kbxdecoder.py firmware.bin --key https://example.com/keys.txt --key ./extra.txt --key <KEY>

Scan all files in the script's directory:

python kbxdecoder.py --all --key keys.txt

Output

Example output for found keyboxes:

<source>_keybox_raw.bin                        - Raw encrypted keybox blob
<source>_keybox_<device_id>_l1_decrypted.bin   - Decrypted payload
<source>_keybox_<device_id>_l1_wvkeybox.bin    - 128-byte Widevine keybox
<source>_keybox_<device_id>_l1.wvd             - WVD file
manifest.json                                  - JSON logs results

The <device_id> comes from the decrypted keybox.

Tested AES

Each key gets tested against:

ECB
CBC with zero IV
CBC with first 16 bytes of ciphertext as IV

CBC key pairs

To supply a specific key + IV for CBC, format lines in your key file as:

AABBCCDD...,00112233...

WVD generation

WVD generation requires pywidevine:

client_id.bin      - Provisioned ClientIdentification
private_key.pem    - Device RSA private key

Provision the extracted keybox. Raw keyboxes contain the Device AES Key and provisioning token. RSA keypair is generated. Once you have both files, run with --extract-wvd to generate the .wvd.

Key cache

Keys are saved as keys.txt in the output directory. Force a re-download with --force-update-keys.

All flags

Flag	Description
`files`	One or more binary files to scan
`--all`	Scan all files in the script's directory
`-o`, `--output-dir`	Directory to save all output files and key cache
`--key HEX\|FILE\|URL`	AES key source - hex string, local file, or URL
`--only-custom`	Only use keys passed via `--key`, skip URL cache
`--force-update-keys`	Force re-download of URL key sources even if cache is fresh
`--no-decrypt`	Skip decryption, only extract raw keybox blobs
`--all-matches`	Save output for every key/mode combination that succeeds
`--extract-wvd`	Attempt `.wvd` generation
`--no-hexview`	Suppress hex dump output
`--verbose-filters`	Print the reason for rejections
`-q`, `--quiet`	Suppress all output except errors

prxtractor

Extracts PlayReady passphrases from libplayready.so and decrypts the associated bgroupcert.dat / zgpriv.dat files. Also unwraps zgpriv_protected.dat for OEMs that reused the PlayReady Porting Kit default Transient/Intermediate keys.

How?

Two paths, picked per-file by inspecting the first 8 bytes:

Files starting with Salted__ are decrypted with OpenSSL EVP_BytesToKey (MD5 KDF, AES-256-CBC) using a passphrase. The passphrase is either auto-detected by scanning the supplied .so for null-separated pszBasePhrase + pszAdditionalPhrase strings, or supplied with --phrase.
Anything else is treated as an RFC 3394 AES-Key-Wrap blob and unwrapped with a KEK derived from the PlayReady Porting Kit hardcoded Transient Key (8B22...427F) and Intermediate Key (9CE9...E136) via AES-CMAC KDF in counter mode. First 32 bytes of the unwrapped material are written as zgpriv.

Phrase list

Candidate passphrases live in phrases.txt next to the script, one per line. The script auto-loads this file and uses it for two things:

Hinting candidates when the .so contains a known base prefix but the additional tail can't be auto-paired
The --list-phrases output

Add new phrases as you discover them from strings/binwalk of new libplayready.so builds. Point --phrases-file at a different file if you keep multiple lists.

Usage

Scan a library and decrypt both .dat files:

python prxtractor.py libplayready.so.0 bgroupcert.dat zgpriv.dat

Unwrap a zgpriv_protected.dat (no .so needed, no passphrase needed):

python prxtractor.py "" zgpriv_protected.dat

Override the auto-detected passphrase (useful when the .so is stripped):

python prxtractor.py libplayready.so.0 bgroupcert.dat zgpriv.dat --phrase <phrase>

List the known Panasonic passphrase candidates:

python prxtractor.py --list-phrases

All flags

Flag	Description
`so_path`	Path to `libplayready.so.0` (or any binary) to scan for passphrases. Optional if you only need to unwrap a `zgpriv_protected.dat`.
`dat_files`	One or more `.dat` files. Format is auto-detected per file.
`--phrase`	Override the auto-detected passphrase.
`--phrases-file`	Path to a candidate passphrase list (default: `phrases.txt` next to the script).
`--list-phrases`	Print the candidates from the phrase file and exit.