abstracted Rng

Cargo.lock

@@ -8,6 +8,12 @@ version = "1.2.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "aae1277d39aeec15cb388266ecc24b11c80469deae6067e17a1a7aa9e5c1f234"
 
+[[package]]
+name = "bitflags"
+version = "2.11.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "c4512299f36f043ab09a583e57bceb5a5aab7a73db1805848e8fef3c9e8c78b3"
+
 [[package]]
 name = "cfg-if"
 version = "1.0.4"
@@ -23,6 +29,12 @@ dependencies = [
  "cfg-if",
 ]
 
+[[package]]
+name = "diceware_wordlists"
+version = "1.2.3"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "5f9b52b69d268c7a2bc582e3aec5cdfa43ac91cef4fe6b6751b02da2b43d6166"
+
 [[package]]
 name = "getrandom"
 version = "0.2.17"
@@ -82,6 +94,8 @@ checksum = "f8ca58f447f06ed17d5fc4043ce1b10dd205e060fb3ce5b979b8ed8e59ff3f79"
 name = "password-generator"
 version = "0.1.0"
 dependencies = [
+ "bitflags",
+ "diceware_wordlists",
  "rand",
  "webster",
 ]

Cargo.toml

@@ -6,3 +6,5 @@ edition = "2024"
 [dependencies]
 rand = "0.8"
 webster = "0.3.0"
+bitflags = "2"
+diceware_wordlists = "1.2.3"

src/main.rs

@@ -1,74 +1,335 @@
 use rand::Rng;
+use rand::RngCore;
 use rand::rngs::OsRng;
-use std::fs;
+use rand::SeedableRng;
+use rand::rngs::StdRng;
 
-// alphanumeric only
-const CHARSET: &[u8] =
-    b"ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789";
+// use std::fs;
+use rand::prelude::SliceRandom;
+use bitflags::bitflags;
 
-// excludes backtick(`), quote('), dquote("), slash(/), bslash(\)
-const CHARSET_WITH_SYMBOLS: &[u8] = 
-    b"ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789!@#$%^&*()[]{}|?<>,.-_=+~";
+use diceware_wordlists::Wordlist;
 
+const UPPERCASE_CHARS: &[u8] = b"ABCDEFGHIJKLMNOPQRSTUVWXYZ";
+const LOWERCASE_CHARS: &[u8] = b"abcdefghijklmnopqrstuvwxyz";
+const NUMBER_CHARS: &[u8] = b"0123456789";
+// still exclude symbols that can trip up URLs and CLIs
+// excludes: backtick(`), quote('), dquote("), slash(/), bslash(\)
+//           pipe (|), arrows (<), (>), brackets ([), (]), ({), (})
+const SYMBOL_CHARS: &[u8] = b"!@#$%^&*()?,.-_=+~"; // 18 chars
 
-fn random_ascii(rng: &mut OsRng, char_set: &[u8]) -> u8 {
+// saves some boilerplate manual bitflags and less clunky to use
+bitflags! {
+    struct CharSet: u8 {
+        const UPPERCASE = 0b0001;
+        const LOWERCASE = 0b0010;
+        const NUMBERS   = 0b0100;
+        const SYMBOLS   = 0b1000;
+    }
+}
+
+struct PasswordConfig {
+    len: usize,
+    charset: CharSet,
+}
+
+// aka password policy
+struct PassphraseConfig {
+    phrases: usize,
+    separator: char,
+    capitalize: bool,
+    number: bool,
+    symbol: bool
+}
+
+impl PassphraseConfig {
+    fn most_secure_passphrase_config() -> Self {
+        Self {
+            phrases: 4,
+            separator: '-',
+            capitalize: true,
+            number: true,
+            symbol: true
+        }
+    }
+}
+
+fn validate_password(password: &str, policy: &PasswordConfig) -> bool {
+    // todo: perf improvement, iterate through string only once
+    if policy.charset.contains(CharSet::UPPERCASE) {
+        let mut found_upper = false;
+        for c in password.chars() {
+            if UPPERCASE_CHARS.contains(&(c as u8)) {
+                found_upper = true;
+                break;
+            }
+        }    
+        if !found_upper {
+            return false;
+        }
+    }
+    if policy.charset.contains(CharSet::LOWERCASE) {
+        let mut found_lower = false;
+        for c in password.chars() {
+            if LOWERCASE_CHARS.contains(&(c as u8)) {
+                found_lower = true;
+                break;
+            }
+        }    
+        if !found_lower {
+            return false;
+        }
+    }
+    if policy.charset.contains(CharSet::NUMBERS) {
+        let mut found_number = false;
+        for c in password.chars() {
+            if NUMBER_CHARS.contains(&(c as u8)) {
+                found_number = true;
+                break;
+            }
+        }    
+        if !found_number {
+            return false;
+        }
+    }
+    if policy.charset.contains(CharSet::SYMBOLS) {
+        let mut found_symbol = false;
+        for c in password.chars() {
+            if SYMBOL_CHARS.contains(&(c as u8)) {
+                found_symbol = true;
+                break;
+            }
+        }    
+        if !found_symbol {
+            return false;
+        }
+    }
+    true
+}
+
+fn validate_passphrase(password: &str, policy: &PassphraseConfig) -> bool {
+    let mut symbol = false;
+    let mut number = false;
+    let mut lower = false;
+    let mut upper = false;
+    for c in password.chars() {
+        if UPPERCASE_CHARS.contains(&(c as u8)) {
+            upper = true;
+            continue;
+        }
+        if LOWERCASE_CHARS.contains(&(c as u8)) {
+            lower = true;
+            continue;
+        }
+        if SYMBOL_CHARS.contains(&(c as u8)) {
+            symbol = true;
+            continue;
+        }
+        if NUMBER_CHARS.contains(&(c as u8)) {
+            number = true;
+            continue;
+        }
+    }
+    (policy.capitalize || upper) &&
+        (policy.symbol || symbol) &&
+        (policy.number || number) &&
+        (lower) // all passwords need lower chars
+}
+
+impl Default for PasswordConfig {
+    fn default() -> Self {
+        let charset = CharSet::UPPERCASE |
+                      CharSet::LOWERCASE |
+                      CharSet::NUMBERS;
+        Self {
+            len: 16,
+            charset 
+        }
+    }
+}
+
+fn random_ascii(rng: &mut MyRng, char_set: &[u8]) -> u8 {
     // NOT IDEAL FOR CRYPTO/password generation 
     // let range_max = 122; // 125 - 33
     // let n = rng.next_u32() % range_max; // slightly modulo biased
     // (n + 33) as u8
 
+    // todo: validate this fact
     // this function under the hood creates a zone that divides evenly
     // into the range so we don't encounter modulo bias
-    let index = rng.gen_range(0..char_set.len()); 
-    char_set[index]
+    *char_set.choose(rng.as_rng()).expect("char_set should not be empty")
 }
 
-fn generate_random_chars(len: usize, symbols: bool, rng: &mut OsRng) -> String {
+// returns error if generated password doesn't contain the necessary chars
+// to satisfy the policy
+// let the caller decide what to do
+fn generate_password(config: &PasswordConfig, rng: &mut MyRng) -> Result<String,String> {
+    if config.len < 8 {
+        return Err(String::from("Password length not long enough"));
+    }
     // String is a thin wrapper around Vec<u8> so we can expect re-allocation
     // and copy only when we exceed the capacity of the vector
     // However, with_capacity will perform only one allocation anyway
-    let mut res = String::with_capacity(len);
-    let char_set = if symbols { CHARSET_WITH_SYMBOLS } else { CHARSET };
-    for _ in 0..len {
-        res.push(random_ascii(rng, char_set) as char);
+    let mut res = String::with_capacity(config.len);
+    let mut chars : Vec<u8> = Vec::new();
+    if config.charset.contains(CharSet::UPPERCASE) {
+        chars.extend_from_slice(UPPERCASE_CHARS); // 26 chars
     }
-    res
+    if config.charset.contains(CharSet::LOWERCASE) {
+        chars.extend_from_slice(LOWERCASE_CHARS); // 26 chars
+    }
+    if config.charset.contains(CharSet::NUMBERS) {
+        chars.extend_from_slice(NUMBER_CHARS); // 10 chars
+    }
+    if config.charset.contains(CharSet::SYMBOLS) {
+        chars.extend_from_slice(SYMBOL_CHARS); // 18 chars
+    }
+
+    // How many bits of entropy?
+    // max_chars_available -> 80
+    // log2(possible_symbols^len) yields bits of entropy
+    // so say len is 20 and we use all 80 chars
+    // that's log2(80^20) -> 126.4 bits of entropy per password
+    for _ in 0..config.len {
+        res.push(random_ascii(rng, &chars) as char);
+    }
+    if !validate_password(&res, config) {
+        return Err("Password does not satisfy supplied config/policy".to_string());
+    }
+    Ok(res)
 }
 
+// todo: haven't updated bits of entropy comments since adding capitalization,
+//       symbols, numbers and changing to diceware wordlist
+//
+// How many bits of entropy?
+// log2(104078) -> 16.6 bits per word derived from word_list size
+// log2(18) -> 4.1 bits for separator assuming seporator is one of our SYMBOLs
+// total 70.5 bits of entropy
+//
 // we can take a slice of "words" because we don't care about:
 // capacity, mutation (push/pop), allocation strategy
-fn generate_passphrase(phrase_count: usize,
-        word_list: &[&str], rng: &mut OsRng) -> String {
-    let mut res = String::with_capacity(phrase_count*2); // save a couple allocations
-    for _ in 0..phrase_count {
-        res.push_str(random_word(word_list, rng));
+fn generate_passphrase(config: &PassphraseConfig,
+        word_list: &[&str], rng: &mut MyRng) -> Result<String,String> {
+    if config.phrases < 2 {
+        return Err("Phrase count too low".to_string());
     }
-    res
+    let mut res = String::new(); // don't bother saving allocations with a bad guess 
+    for i in 0..config.phrases {
+        let word = random_word(word_list, rng);
+        if config.capitalize {
+            let mut word_chars = word.chars();
+            let capitalized_word : String = word_chars.next().unwrap()
+                .to_uppercase().chain(word_chars).collect();
+            res.push_str(&capitalized_word);
+        } else {
+            res.push_str(word);
+        }
+        if i < config.phrases - 1 {
+            res.push(config.separator);
+        }
+    }
+    if config.symbol {
+        res.push(random_ascii(rng, SYMBOL_CHARS) as char)
+    }
+    if config.number {
+        res.push(random_ascii(rng, NUMBER_CHARS) as char)
+    }
+    Ok(res)
 }
 
-fn random_word<'a>(word_list: &'a [&'a str], rng: &mut OsRng) -> &'a str {
-    word_list[rng.gen_range(0..word_list.len())]
+fn random_word<'a>(word_list: &[&'a str], rng: &mut MyRng) -> &'a str {
+    word_list[rng.as_rng().gen_range(0..word_list.len())]
 }
 
+fn get_word_list() -> &'static [&'static str] {
+    // platform dependent word list
+    // let content = fs::read_to_string("/usr/share/dict/words")
+    //     .expect("could not read dictionary");
+    // filter words here for only ascii
+    // word_list length is 104078
+    // let word_list: Vec<&str> = content.lines().filter(|w|
+    //     w.is_ascii()).collect();
+
+    // platform independent word list
+    // this list doesn't seem to contain non-ascii or even punctuation
+    Wordlist::get_list(&Wordlist::EffLong)
+}
+
+enum Determinism {
+    Predictable,
+    #[allow(dead_code)]
+    Random,
+}
+
+enum MyRng { Std(StdRng), Os(OsRng) }
+
+impl MyRng {
+    // todo: understand RngCore
+    fn as_rng(&mut self) -> &mut dyn RngCore {
+        match self {
+            MyRng::Std(r) => r,
+            MyRng::Os(r) => r,
+        }
+    }
+}
+
+fn get_rng(d: &Determinism) -> MyRng {
+    match d {
+        // when we are looking for Reproducability we can use the same seed value
+        // to create our Rng. Because both StdRng and OsRng implement Rng
+        // we make our methods slightly more generic and this works just fine
+        // comment out when we need random version!
+        Determinism::Predictable => MyRng::Std(StdRng::seed_from_u64(42)),
+        // we definitely want to use the OS for randomness
+        // doesn't require seeding because it is backed by CSPRNG, backed by OS
+        // the OS maintains it's own entropy from mouse movement, hardware noise,etc
+        // we pass it through the functions to avoid a potential syscall
+        Determinism::Random => MyRng::Os(OsRng),
+    }
+}
 
 fn main() {
-    let content = fs::read_to_string("/usr/share/dict/words")
-        .expect("could not read dictionary");
+    let word_list = get_word_list();
 
-    // filter words here for only ascii
-    let word_list: Vec<&str> = content.lines().filter(|w|
-        w.is_ascii()).collect();
+    let mut rng = get_rng(&Determinism::Predictable);
 
-    // we definitely want to use the OS for randomness
-    // doesn't require seeding because it is backed by CSPRNG, backed by OS
-    // the OS maintains it's own entropy from mouse movement, hardware noise,etc
-    // we pass it through the functions to avoid a potential syscall
-    let mut rng = OsRng;
-    let chars = generate_random_chars(20, true, &mut rng);
-    println!("generated random chars: {}", chars.as_str());
+    // todo: add actual tests
+    // password with alphnumeric
+    let password_res = generate_password(&Default::default(), &mut rng);
+    if let Err(e) = password_res {
+        println!("Error during first password gen: {}", e);
+    } else {
+        let password = password_res.unwrap();
+        println!("generated random chars (alphanumeric): {}", password.as_str());
+        // this will fail intermittently since we haven't added any guarantees
+        // that specified charsets are generated.
+        assert!(validate_password(&password, &Default::default()));
+    }
 
-    let phrase = generate_passphrase(3, &word_list, &mut rng);
-    println!("generated passphrase: {}", phrase.as_str());
+    // password with alphanumeric + symbols
+    let config = PasswordConfig {
+        charset: CharSet::SYMBOLS,
+        ..Default::default()
+    };
+    let password_res2 = generate_password(&config, &mut rng);
+    // this will fail intermittently since we haven't added any guarantees
+    // that specified charsets are generated.
+    // Also note: we just unwrap here. We should handle like above
+    assert!(validate_password(password_res2.as_ref().unwrap(), &config));
+    println!("generated random chars (with symbols): {}",
+        password_res2.unwrap().as_str());
 
-    println!("words in dictionary: {}", word_list.len());
+    let passphrase_config = PassphraseConfig {
+        phrases: 3,
+        separator: '-',
+        capitalize: true,
+        number: true,
+        symbol: true,
+    };
+    let passphrase_res = generate_passphrase(&passphrase_config, &word_list, &mut rng);
+    // Also note: we just unwrap here. We should handle like above
+    assert!(validate_passphrase(passphrase_res.as_ref().unwrap(),
+        &PassphraseConfig::most_secure_passphrase_config()));
+    println!("generated passphrase: {}", passphrase_res.as_ref().unwrap());
 }