[mg3d] feat #3: implement migu 3d decipher with improved key guessing

.idea/lib_um_crypto.iml

@@ -15,6 +15,7 @@
       <sourceFolder url="file://$MODULE_DIR$/um_crypto/xmly/src" isTestSource="false" />
       <sourceFolder url="file://$MODULE_DIR$/um_crypto/xiami/src" isTestSource="false" />
       <sourceFolder url="file://$MODULE_DIR$/um_crypto/qtfm/src" isTestSource="false" />
+      <sourceFolder url="file://$MODULE_DIR$/um_crypto/mg3d/src" isTestSource="false" />
       <excludeFolder url="file://$MODULE_DIR$/target" />
       <excludeFolder url="file://$MODULE_DIR$/um_wasm_loader/dist" />
       <excludeFolder url="file://$MODULE_DIR$/um_wasm_loader/pkg" />

Cargo.lock

@@ -535,6 +535,7 @@ dependencies = [
  "umc_joox",
  "umc_kgm",
  "umc_kuwo",
+ "umc_mg3d",
  "umc_ncm",
  "umc_qmc",
  "umc_qtfm",
@@ -600,6 +601,15 @@ dependencies = [
  "umc_utils",
 ]
 
+[[package]]
+name = "umc_mg3d"
+version = "0.1.0"
+dependencies = [
+ "hex",
+ "thiserror",
+ "umc_utils",
+]
+
 [[package]]
 name = "umc_ncm"
 version = "0.1.0"

um_cli/Cargo.toml

@@ -10,6 +10,7 @@ hex = "0.4.3"
 umc_joox = { path = "../um_crypto/joox" }
 umc_kgm = { path = "../um_crypto/kgm" }
 umc_kuwo = { path = "../um_crypto/kuwo" }
+umc_mg3d = { path = "../um_crypto/mg3d" }
 umc_ncm = { path = "../um_crypto/ncm" }
 umc_qmc = { path = "../um_crypto/qmc" }
 umc_qtfm = { path = "../um_crypto/qtfm" }

um_cli/src/cmd/mg3d.rs

@@ -0,0 +1,70 @@
+use crate::Cli;
+use anyhow::{bail, Result};
+use clap::Args;
+use std::fs::File;
+use std::io::{BufReader, BufWriter, Read, Seek, SeekFrom, Write};
+use std::path::PathBuf;
+use umc_mg3d::{guess_key, Decipher};
+
+/// Decrypt a mg3d file (Migu 3D Audio)
+#[derive(Args)]
+pub struct ArgsMigu3D {
+    /// Path to output file, e.g. /export/Music/song.wav
+    #[clap(short, long)]
+    output: PathBuf,
+
+    /// Path to input file, e.g. /export/Music/song.mg3d
+    #[arg(name = "input")]
+    input: PathBuf,
+
+    /// File key (androidFileKey/iosFileKey). Leave empty to guess the key.
+    #[clap(short = 'k', long = "file-key")]
+    file_key: Option<String>,
+}
+
+impl ArgsMigu3D {
+    pub fn run(&self, cli: &Cli) -> Result<i32> {
+        let mut reader = BufReader::with_capacity(cli.buffer_size, File::open(&self.input)?);
+        let mut writer = BufWriter::with_capacity(cli.buffer_size, File::create(&self.output)?);
+
+        let decipher = self.make_decipher(&mut reader)?;
+        if cli.verbose {
+            println!("final key: {}", hex::encode(decipher.get_key()));
+        }
+
+        let mut offset = 0usize;
+        let mut buffer = vec![0u8; cli.buffer_size];
+        loop {
+            let n = reader.read(&mut buffer[..])?;
+            if n == 0 {
+                break;
+            }
+            decipher.decrypt(&mut buffer[..n], offset);
+            writer.write_all(&buffer[..n])?;
+            offset += n;
+        }
+
+        Ok(0)
+    }
+
+    fn make_decipher<T>(&self, reader: &mut T) -> Result<Decipher>
+    where
+        T: Read + Seek + ?Sized,
+    {
+        let decipher = match &self.file_key {
+            None => {
+                let mut buffer = [0u8; 0x100];
+                reader.read_exact(&mut buffer)?;
+                reader.seek(SeekFrom::Current(-0x100))?;
+                let key = match guess_key(&buffer) {
+                    Some(key) => key,
+                    None => bail!("failed to guess a valid key"),
+                };
+                Decipher::new_from_final_key(&key)?
+            }
+            Some(key) => Decipher::new_from_file_key(key)?,
+        };
+
+        Ok(decipher)
+    }
+}

um_cli/src/cmd/mod.rs

@@ -2,6 +2,7 @@ use clap::Subcommand;
 
 pub mod joox;
 pub mod kgm;
+pub mod mg3d;
 pub mod ncm;
 pub mod qmc1;
 pub mod qmc2;
@@ -11,20 +12,29 @@ pub mod xmly;
 
 #[derive(Subcommand)]
 pub enum Commands {
+    #[command(name = "ncm")]
+    NCM(ncm::ArgsNCM),
+
+    #[command(name = "kgm")]
+    KGM(kgm::ArgsKGM),
+
+    #[command(name = "mg3d")]
+    Migu3D(mg3d::ArgsMigu3D),
+
+    #[command(name = "joox")]
+    JOOX(joox::ArgsJoox),
+
     #[command(name = "qmc1")]
     QMCv1(qmc1::ArgsQMCv1),
     #[command(name = "qmc2")]
     QMCv2(qmc2::ArgsQMCv2),
-    #[command(name = "ncm")]
-    NCM(ncm::ArgsNCM),
-    #[command(name = "kgm")]
-    KGM(kgm::ArgsKGM),
-    #[command(name = "joox")]
-    JOOX(joox::ArgsJoox),
-    #[command(name = "xmly")]
-    XMLY(xmly::ArgsXimalaya),
-    #[command(name = "xiami")]
-    Xiami(xiami::ArgsXiami),
+
     #[command(name = "qtfm")]
     QTFM(qtfm::ArgsQingTingFM),
+
+    #[command(name = "xiami")]
+    Xiami(xiami::ArgsXiami),
+
+    #[command(name = "xmly")]
+    XMLY(xmly::ArgsXimalaya),
 }

um_cli/src/main.rs

@@ -27,14 +27,15 @@ pub struct Cli {
 
 fn run_command(cli: &Cli) -> Result<i32> {
     match &cli.command {
+        Some(Commands::JOOX(cmd)) => cmd.run(&cli),
+        Some(Commands::KGM(cmd)) => cmd.run(&cli),
+        Some(Commands::Migu3D(cmd)) => cmd.run(&cli),
+        Some(Commands::NCM(cmd)) => cmd.run(&cli),
         Some(Commands::QMCv1(cmd)) => cmd.run(&cli),
         Some(Commands::QMCv2(cmd)) => cmd.run(&cli),
-        Some(Commands::NCM(cmd)) => cmd.run(&cli),
-        Some(Commands::KGM(cmd)) => cmd.run(&cli),
-        Some(Commands::JOOX(cmd)) => cmd.run(&cli),
-        Some(Commands::XMLY(cmd)) => cmd.run(&cli),
-        Some(Commands::Xiami(cmd)) => cmd.run(&cli),
         Some(Commands::QTFM(cmd)) => cmd.run(&cli),
+        Some(Commands::Xiami(cmd)) => cmd.run(&cli),
+        Some(Commands::XMLY(cmd)) => cmd.run(&cli),
         None => {
             // https://github.com/clap-rs/clap/issues/3857#issuecomment-1161796261
             todo!("implement a sensible default command, similar to um/cli");

um_crypto/mg3d/Cargo.toml

@@ -0,0 +1,9 @@
+[package]
+name = "umc_mg3d"
+version = "0.1.0"
+edition = "2021"
+
+[dependencies]
+hex = "0.4.3"
+thiserror = "1.0.63"
+umc_utils = { path = "../utils" }

um_crypto/mg3d/Readme.MD

@@ -0,0 +1,29 @@
+# 咪咕 3D 音乐
+
+整个文件加密，每个文件有独立的密钥。
+
+```py
+mg3d_file_key <- md5(b"AC89EC47A70B76F307CB39A0D74BCCB0" + file_key).hex(upper=true)
+```
+
+生成最终的 `mg3d_file_key` 后，每个字节依序减少密钥的字节内容。
+
+## WAV 格式
+
+几乎固定的文件头。
+
+```text
+000:0000  52 49 46 46 ?? ?? ?? ?? 57 41 56 45 66 6D 74 20  RIFF....WAVEfmt 
+000:0010  ?? ?? ?? ?? ?? ?? ?? ?? ?? ?? ?? ?? ?? ?? ?? ??  ................
+000:0020  06 00 18 00 6A 75 6E 6B 34 00 00 00 00 00 00 00  ....junk4.......
+000:0030  00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00  ................
+000:0040  00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00  ................
+000:0050  00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00  ................
+000:0060  64 61 74 61 ?? ?? ?? ?? ?? ?? ?? ?? ?? ?? ?? ??  data............
+```
+
+验证一下前四字节是否解密得到 `RIFF`、以及 0x60 处是否位 `data` 来确定。
+
+## M4A 格式
+
+前 0x1C 字节固定 (其实前 4 字节有可能会变，但他们用的编码器好像不会变)

um_crypto/mg3d/src/guess_m4a.rs

@@ -0,0 +1,73 @@
+use crate::is_valid_password_chr;
+use std::collections::HashMap;
+
+const GUESS_PLAIN_TEXT: [u8; 0x20] = [
+    0x00, 0x00, 0x00, 0x00, 0x66, 0x74, 0x79, 0x70, 0x4D, 0x34, 0x41, 0x20, 0x00, 0x00, 0x00, 0x00,
+    0x4D, 0x34, 0x41, 0x20, 0x6D, 0x70, 0x34, 0x32, 0x69, 0x73, 0x6F, 0x6D, 0x00, 0x00, 0x00, 0x00,
+];
+
+type ByteFreq = HashMap<u8, usize>;
+fn get_highest_freq_item(freq: &ByteFreq) -> u8 {
+    let mut current_item = 0u8;
+    let mut current_count = 0usize;
+
+    for (&item, &count) in freq.iter() {
+        if count > current_count {
+            current_item = item;
+            current_count = count;
+        }
+    }
+
+    current_item
+}
+
+pub fn guess_key(buffer: &[u8]) -> Option<[u8; 0x20]> {
+    if buffer.len() < 0x100 {
+        // buffer too small
+        None?
+    }
+
+    let mut key = [0u8; 0x20];
+    key.copy_from_slice(&buffer[0..0x20]);
+
+    for (k, plain) in key.iter_mut().zip(GUESS_PLAIN_TEXT) {
+        *k = k.wrapping_sub(plain);
+    }
+    if !&key[0x04..0x1C].iter().all(|&k| is_valid_password_chr(k)) {
+        // Includes non-password chr
+        None?
+    }
+
+    let mut password_0x03_freq = ByteFreq::new();
+    let mut password_0x1c_freq = ByteFreq::new();
+    let mut password_0x1d_freq = ByteFreq::new();
+    let mut password_0x1e_freq = ByteFreq::new();
+    let mut password_0x1f_freq = ByteFreq::new();
+
+    let increment_password_freq_count = |freq: &mut ByteFreq, item: u8| {
+        if is_valid_password_chr(item) {
+            freq.entry(item)
+                .and_modify(|counter| *counter += 1)
+                .or_insert(1);
+        }
+    };
+
+    for chunk in buffer[..0x100].chunks(0x20) {
+        increment_password_freq_count(&mut password_0x03_freq, chunk[0x03]);
+        increment_password_freq_count(&mut password_0x1c_freq, chunk[0x1c]);
+        increment_password_freq_count(&mut password_0x1d_freq, chunk[0x1d]);
+        increment_password_freq_count(&mut password_0x1e_freq, chunk[0x1e]);
+        increment_password_freq_count(&mut password_0x1f_freq, chunk[0x1f]);
+    }
+    key[0x03] = get_highest_freq_item(&password_0x03_freq);
+    key[0x1C] = get_highest_freq_item(&password_0x1c_freq);
+    key[0x1D] = get_highest_freq_item(&password_0x1d_freq);
+    key[0x1E] = get_highest_freq_item(&password_0x1e_freq);
+    key[0x1F] = get_highest_freq_item(&password_0x1f_freq);
+
+    if is_valid_password_chr(key[0x03]) && key[0x1c..].iter().all(|&c| is_valid_password_chr(c)) {
+        Some(key)
+    } else {
+        None
+    }
+}

um_crypto/mg3d/src/guess_wav.rs

@@ -0,0 +1,28 @@
+use crate::{is_valid_password_chr, raw_decrypt};
+
+pub fn guess_key(buffer: &[u8]) -> Option<[u8; 0x20]> {
+    if buffer.len() < 0x100 {
+        // buffer too small
+        None?
+    }
+
+    let mut key = [0u8; 0x20];
+    key.copy_from_slice(&buffer[0x40..0x60]);
+    if !key.iter().all(|&k| is_valid_password_chr(k)) {
+        // Not valid password
+        None?
+    }
+
+    let mut test_riff = [0u8; 4];
+    test_riff.copy_from_slice(&buffer[0..4]);
+    raw_decrypt(&mut test_riff, &key, 0x00);
+
+    let mut test_data = [0u8; 4];
+    test_data.copy_from_slice(&buffer[0x60..0x64]);
+    raw_decrypt(&mut test_data, &key, 0x60);
+
+    match (&test_riff, &test_data) {
+        (b"RIFF", b"data") => Some(key),
+        (_, _) => None,
+    }
+}

um_crypto/mg3d/src/lib.rs

@@ -0,0 +1,61 @@
+use thiserror::Error;
+use umc_utils::md5_2;
+
+mod guess_m4a;
+mod guess_wav;
+
+pub use guess_m4a::guess_key as guess_m4a_key;
+pub use guess_wav::guess_key as guess_wav_key;
+
+pub fn guess_key(buffer: &[u8]) -> Option<[u8; 0x20]> {
+    guess_wav_key(buffer).or_else(|| guess_m4a_key(buffer))
+}
+
+fn raw_decrypt<T: AsMut<[u8]> + ?Sized>(buffer: &mut T, key: &[u8; 0x20], offset: usize) {
+    for (b, i) in buffer.as_mut().iter_mut().zip(offset..) {
+        *b = (*b).wrapping_sub(key[i % key.len()]);
+    }
+}
+
+#[derive(Error, Debug)]
+pub enum Migu3dError {
+    #[error("Invalid FileKey")]
+    InvalidFileKey,
+
+    #[error("Convert hash to key error")]
+    ConvertKeyError,
+}
+
+fn is_valid_password_chr(chr: u8) -> bool {
+    matches!(chr, b'0'..=b'9' | b'A'..=b'F')
+}
+
+pub struct Decipher {
+    key: [u8; 0x20],
+}
+
+impl Decipher {
+    /// Init decipher from "file_key" (androidFileKey or iosFileKey)
+    pub fn new_from_file_key(file_key: &str) -> Result<Self, Migu3dError> {
+        let hash = md5_2(b"AC89EC47A70B76F307CB39A0D74BCCB0", file_key.as_bytes());
+        let key = hex::encode_upper(hash);
+        let key = key
+            .as_bytes()
+            .try_into()
+            .map_err(|_| Migu3dError::ConvertKeyError)?;
+        Ok(Self { key })
+    }
+
+    /// Init decipher from "key" (the final hash)
+    pub fn new_from_final_key(key: &[u8; 0x20]) -> Result<Self, Migu3dError> {
+        Ok(Self { key: *key })
+    }
+
+    pub fn decrypt<T: AsMut<[u8]> + ?Sized>(&self, buffer: &mut T, offset: usize) {
+        raw_decrypt(buffer, &self.key, offset)
+    }
+
+    pub fn get_key(&self) -> [u8; 0x20] {
+        self.key
+    }
+}

um_crypto/utils/src/lib.rs

@@ -1,3 +1,3 @@
 pub mod base64;
 mod md5;
-pub use md5::md5;
+pub use md5::{md5, md5_2};

um_crypto/utils/src/md5.rs

@@ -3,3 +3,10 @@ use md5::{Digest, Md5};
 pub fn md5<T: AsRef<[u8]>>(buffer: T) -> [u8; 16] {
     Md5::digest(buffer).into()
 }
+
+pub fn md5_2<T1: AsRef<[u8]>, T2: AsRef<[u8]>>(buffer1: T1, buffer2: T2) -> [u8; 16] {
+    let mut md5_digest = Md5::default();
+    md5_digest.update(buffer1);
+    md5_digest.update(buffer2);
+    md5_digest.finalize().into()
+}