Add bdf read and write implementation

src/lib/crypt.rs

@@ -1,4 +1,4 @@
-use crate::lib::hash::{sha_checksum, PassKey};
+use crate::lib::hash::sha_checksum;
 use cfb_mode::stream_cipher::{NewStreamCipher, StreamCipher};
 use cfb_mode::Cfb;
 use des::Des;
@@ -33,7 +33,7 @@ pub fn decrypt_data(data: &[u8], key: &[u8]) -> Vec<u8> {
 /// Decrypts data using a dictionary
 pub fn decrypt_with_dictionary(
     data: &[u8],
-    dict: Vec<PassKey>,
+    dict: Vec<(&String, Vec<u8>)>,
     checksum: &[u8],
 ) -> Option<Vec<u8>> {
     let decrypted = Mutex::<Option<Vec<u8>>>::new(None);

src/lib/rainbowutils.rs

@@ -6,6 +6,7 @@ use std::convert::{TryFrom, TryInto};
 use std::fs::File;
 use std::io::{BufReader, BufWriter, Read, Write};
 use std::io::{Error, ErrorKind};
+use xz2::read::{XzDecoder, XzEncoder};
 
 pub const LZMA: &str = "lzma";
 
@@ -18,8 +19,9 @@ pub const ENTRIES_PER_CHUNK: u32 = 100_000;
 
 pub struct BDFReader {
     reader: BufReader<File>,
-    metadata: Option<MetaChunk>,
-    lookup_table: Option<HashLookupTable>,
+    pub metadata: Option<MetaChunk>,
+    pub lookup_table: Option<HashLookupTable>,
+    compressed: bool,
 }
 
 pub struct BDFWriter {
@@ -28,6 +30,7 @@ pub struct BDFWriter {
     lookup_table: HashLookupTable,
     data_entries: Vec<DataEntry>,
     head_written: bool,
+    compressed: bool,
 }
 
 #[derive(Debug, Clone)]
@@ -40,10 +43,10 @@ pub struct GenericChunk {
 
 #[derive(Debug, Clone)]
 pub struct MetaChunk {
-    chunk_count: u32,
+    pub chunk_count: u32,
     entries_per_chunk: u32,
-    entry_count: u64,
-    compression_method: Option<String>,
+    pub entry_count: u64,
+    pub compression_method: Option<String>,
 }
 
 #[derive(Debug, Clone)]
@@ -60,18 +63,19 @@ pub struct HashEntry {
 
 #[derive(Debug, Clone)]
 pub struct DataEntry {
-    plain: String,
+    pub plain: String,
     hashes: HashMap<String, Vec<u8>>,
 }
 
 impl BDFWriter {
     pub fn new(writer: BufWriter<File>, entry_count: u64, compress: bool) -> Self {
         Self {
-            metadata: MetaChunk::new(entry_count, 0, compress),
+            metadata: MetaChunk::new(entry_count, ENTRIES_PER_CHUNK, compress),
             lookup_table: HashLookupTable::new(HashMap::new()),
             data_entries: Vec::new(),
             writer,
             head_written: false,
+            compressed: compress,
         }
     }
 
@@ -115,6 +119,9 @@ impl BDFWriter {
         }
         let mut data_chunk =
             GenericChunk::from_data_entries(&self.data_entries, &self.lookup_table);
+        if self.compressed {
+            data_chunk.compress()?;
+        }
         let data = data_chunk.serialize();
         self.writer.write(data.as_slice())?;
         self.data_entries = Vec::new();
@@ -133,6 +140,7 @@ impl BDFReader {
             metadata: None,
             lookup_table: None,
             reader,
+            compressed: false,
         }
     }
 
@@ -142,6 +150,16 @@ impl BDFReader {
             return Err(Error::new(ErrorKind::InvalidData, "invalid BDF Header"));
         }
         let meta_chunk: MetaChunk = self.next_chunk()?.try_into()?;
+        if let Some(method) = &meta_chunk.compression_method {
+            if *method == LZMA.to_string() {
+                self.compressed = true;
+            } else {
+                return Err(Error::new(
+                    ErrorKind::Other,
+                    "unsupported compression method",
+                ));
+            }
+        }
         self.metadata = Some(meta_chunk);
 
         if let Some(chunk) = &self.metadata {
@@ -195,13 +213,17 @@ impl BDFReader {
         let mut crc_raw = [0u8; 4];
         let _ = self.reader.read_exact(&mut crc_raw)?;
         let crc = BigEndian::read_u32(&mut crc_raw);
-
-        Ok(GenericChunk {
+        let mut gen_chunk = GenericChunk {
             length,
             name,
             data,
             crc,
-        })
+        };
+        if gen_chunk.name == DTBL_CHUNK_NAME.to_string() && self.compressed {
+            gen_chunk.decompress()?;
+        }
+
+        Ok(gen_chunk)
     }
 }
 
@@ -227,9 +249,12 @@ impl GenericChunk {
         &mut self,
         lookup_table: &HashLookupTable,
     ) -> Result<Vec<DataEntry>, Error> {
+        if self.name == HTBL_CHUNK_NAME.to_string() {
+            return Err(Error::new(ErrorKind::Other, "this is not a data chunk"));
+        }
         let mut entries: Vec<DataEntry> = Vec::new();
         let mut position = 0;
-        while self.data.len() > position {
+        while self.data.len() > (position + 8) {
             let entry_length_raw = &self.data[position..position + 4];
             position += 4;
             let entry_length = BigEndian::read_u32(entry_length_raw);
@@ -240,7 +265,15 @@ impl GenericChunk {
             let pw_plain_raw = &self.data[position..position + pw_length as usize];
             position += pw_length as usize;
             let pw_plain = String::from_utf8(pw_plain_raw.to_vec())
-                .expect("failed to parse plain password string");
+                .map_err(|err| {
+                    format!(
+                        "failed to parse plain password string ({}-{}): {:?}",
+                        position,
+                        position + pw_length as usize,
+                        err
+                    )
+                })
+                .unwrap();
             let mut hash_values: HashMap<String, Vec<u8>> = HashMap::new();
             while position < entry_end {
                 let entry_id_raw = &self.data[position..position + 4];
@@ -283,6 +316,35 @@ impl GenericChunk {
             crc: crc_sum,
         }
     }
+
+    pub fn compress(&mut self) -> Result<(), Error> {
+        let data = self.data.as_slice();
+        let mut compressor = XzEncoder::new(data, 6);
+        let mut compressed: Vec<u8> = Vec::new();
+        compressor.read_to_end(&mut compressed)?;
+        self.length = compressed.len() as u32;
+        self.data = compressed;
+
+        Ok(())
+    }
+
+    pub fn decompress(&mut self) -> Result<(), Error> {
+        let data = self.data.as_slice();
+        let mut decompressor = XzDecoder::new(data);
+        let mut decompressed: Vec<u8> = Vec::new();
+        decompressor.read_to_end(&mut decompressed)?;
+        let crc = crc32::checksum_ieee(decompressed.as_slice());
+        if crc != self.crc {
+            return Err(Error::new(
+                ErrorKind::InvalidData,
+                "the crc doesn't match the decrypted data",
+            ));
+        }
+        self.length = decompressed.len() as u32;
+        self.data = decompressed;
+
+        Ok(())
+    }
 }
 
 impl From<&MetaChunk> for GenericChunk {
@@ -321,7 +383,7 @@ impl MetaChunk {
         } else {
             None
         };
-        let chunk_count = (entry_count as f32 / entries_per_chunk as f32).ceil() as u32;
+        let chunk_count = (entry_count as f64 / entries_per_chunk as f64).ceil() as u32;
 
         Self {
             chunk_count,
@@ -494,6 +556,11 @@ impl DataEntry {
         self.hashes.insert(name, value);
     }
 
+    /// Returns the hash value for a given name of a hash function
+    pub fn get_hash_value(&self, name: String) -> Option<&Vec<u8>> {
+        self.hashes.get(&name)
+    }
+
     /// Serializes the entry to a vector of bytes
     pub fn serialize(&self, lookup_table: HashLookupTable) -> Vec<u8> {
         let mut pw_plain_raw = self.plain.clone().into_bytes();

src/main.rs

@@ -3,8 +3,8 @@ pub mod lib;
 use crate::lib::crypt::{
     decrypt_brute_brute_force, decrypt_data, decrypt_with_dictionary, encrypt_data,
 };
-use crate::lib::hash::{create_key, sha_checksum, PassKey};
-use crate::lib::rainbowutils::{BDFWriter, DataEntry, HashEntry};
+use crate::lib::hash::{create_key, sha256, sha_checksum};
+use crate::lib::rainbowutils::{BDFReader, BDFWriter, DataEntry, HashEntry, HashLookupTable};
 use pbr::ProgressBar;
 use rayon::prelude::*;
 use rayon::str;
@@ -12,9 +12,10 @@ use regex::Regex;
 use rpassword;
 use rpassword::read_password_from_tty;
 use spinners::{Spinner, Spinners};
+use std::collections::HashMap;
 use std::fs;
 use std::fs::File;
-use std::io::{BufRead, BufReader, BufWriter, Write};
+use std::io::{BufReader, BufWriter};
 use std::sync::mpsc::sync_channel;
 use std::thread;
 use std::time::Duration;
@@ -85,8 +86,12 @@ struct CreateDictionary {
     input: String,
 
     /// The output dictionary file
-    #[structopt(short = "o", long = "output", default_value = "dictionary.tsv")]
+    #[structopt(short = "o", long = "output", default_value = "dictionary.bdf")]
     output: String,
+
+    /// If the dictionary file should be compressed
+    #[structopt(short = "c", long = "compress")]
+    compress: bool,
 }
 
 fn main() {
@@ -160,7 +165,7 @@ fn create_dictionary(_opts: &Opts, args: &CreateDictionary) {
     let input: String = (*args.input).parse().unwrap();
     // TODO: Some form of removing duplicates (without itertools)
     let fout = File::create(args.output.clone()).unwrap();
-    let mut writer = BufWriter::new(fout);
+    let writer = BufWriter::new(fout);
     let handle;
     {
         let content = fs::read_to_string(input).expect("Failed to read content");
@@ -169,24 +174,28 @@ fn create_dictionary(_opts: &Opts, args: &CreateDictionary) {
         let mut pb = ProgressBar::new(entry_count);
         pb.set_max_refresh_rate(Some(Duration::from_millis(200)));
         let (rx, tx) = sync_channel::<DataEntry>(100_000_000);
-        let mut bdf_file = BDFWriter::new(writer, entry_count, false);
-        bdf_file.add_lookup_entry(HashEntry::new(SHA256.to_string(), 32));
+        let mut bdf_file = BDFWriter::new(writer, entry_count, args.compress);
+        bdf_file
+            .add_lookup_entry(HashEntry::new(SHA256.to_string(), 32))
+            .expect("Failed to add lookup entry");
         handle = thread::spawn(move || {
             for entry in tx {
-                bdf_file.add_data_entry(entry);
+                if let Err(e) = bdf_file.add_data_entry(entry) {
+                    println!("{:?}", e);
+                }
                 pb.inc();
             }
             pb.finish();
-            bdf_file.flush();
+            bdf_file.flush().expect("failed to flush the file data");
             bdf_file
                 .flush_writer()
-                .expect("Failed to flush the file writer.");
+                .expect("failed to flush the file writer");
         });
         let re = Regex::new("[\\x00\\x08\\x0B\\x0C\\x0E-\\x1F\\t\\r\\a\\n]").unwrap();
         lines
             .map(|line| -> String { re.replace_all(line, "").to_string() })
             .map(|pw| -> DataEntry {
-                let key = create_key(pw.replace("\t", "").as_ref());
+                let key = sha256(&pw);
                 let mut data_entry = DataEntry::new(pw);
                 data_entry.add_hash_value(SHA256.to_string(), key);
 
@@ -208,8 +217,6 @@ fn spinner(text: &str) -> Spinner {
     Spinner::new(Spinners::Dots2, text.into())
 }
 
-const LINES_PER_CHUNK: usize = 10000000;
-
 fn decrypt_with_dictionary_file(
     filename: String,
     data: &Vec<u8>,
@@ -218,36 +225,37 @@ fn decrypt_with_dictionary_file(
     let sp = spinner("Reading dictionary...");
     let f = File::open(&filename).expect("Failed to open dictionary file.");
     let reader = BufReader::new(f);
-    let mut pb =
-        ProgressBar::new((get_line_count(&filename) as f64 / LINES_PER_CHUNK as f64).ceil() as u64);
-    let (rx, tx) = sync_channel::<Vec<String>>(10);
+    let mut bdf_file = BDFReader::new(reader);
+    bdf_file
+        .read_metadata()
+        .expect("failed to read the metadata of the file");
+    let mut chunk_count = 0;
+    if let Some(meta) = &bdf_file.metadata {
+        chunk_count = meta.chunk_count;
+    }
+    let mut pb = ProgressBar::new(chunk_count as u64);
+    let (rx, tx) = sync_channel::<Vec<DataEntry>>(10);
     let _handle = thread::spawn(move || {
-        let mut line_vec: Vec<String> = vec![];
-        reader.lines().for_each(|line_result| {
-            if line_vec.len() > LINES_PER_CHUNK {
-                if let Err(_) = rx.send(line_vec.clone()) {}
-                line_vec.clear();
+        let mut lookup_table = HashLookupTable::new(HashMap::new());
+        if let Ok(table) = bdf_file.read_lookup_table() {
+            lookup_table = table.clone();
+        }
+        while let Ok(next_chunk) = &mut bdf_file.next_chunk() {
+            if let Ok(entries) = next_chunk.data_entries(&lookup_table) {
+                if let Err(_) = rx.send(entries) {}
             }
-            match line_result {
-                Ok(line) => line_vec.push(line),
-                Err(err) => eprintln!("Failed with err {}", err),
         }
-        });
-        if let Err(_) = rx.send(line_vec.clone()) {}
-        line_vec.clear();
     });
     sp.stop();
     let mut result_data: Option<Vec<u8>> = None;
-    for lines in tx {
-        let pw_table: Vec<PassKey> = lines
+    for entries in tx {
+        let pw_table: Vec<(&String, Vec<u8>)> = entries
             .par_iter()
-            .map(|line| {
-                let parts: Vec<&str> = line.split("\t").collect::<Vec<&str>>();
-                let pw = parts[0].parse().unwrap();
-                let key_str: String = parts[1].parse().unwrap();
-                let key = base64::decode(&key_str).unwrap();
+            .map(|entry: &DataEntry| {
+                let pw = &entry.plain;
+                let key: &Vec<u8> = entry.get_hash_value(SHA256.to_string()).unwrap();
 
-                (pw, key)
+                (pw, key[0..8].to_vec())
             })
             .collect();
         pb.inc();
@@ -259,9 +267,3 @@ fn decrypt_with_dictionary_file(
     pb.finish();
     result_data
 }
-
-fn get_line_count(fname: &str) -> usize {
-    let f = File::open(fname).expect("Failed to open file to get the line count.");
-    let reader = BufReader::new(f);
-    return reader.lines().count();
-}