Rust

Terms

• • Fat Pointer Contains Address actual data and length

Cargo

Sample file

[package]
name = "hello_world"
version = "0.0.1"
authors = [ "Iain Wiseman iwiseman@bibble.co.nz" ]

Sample commands

 cargo new hello_world --bin
 cargo build
 cargo run

Fundamental Data Types

Primitive types

Cam declare with size of type

 let a:u8 = 123; // unsigned int 8 bits number immutable
 let a:i8 = 123; // signed int 8 bits number immutable
 let mut a:u8 = 123; // unsigned int 8 bits number mutable

Or without e.g.

 let mut c = 123456789 // 32-bit signed i32
 println!("c = {}", c);

Now variable based on OS e.g.

 let z:isize = 123 // signed 64 bit if on 64 bit OS

Decimal

let e:f64 = 2.5 // double-precision, 8 bytes or 64-bits

Char

let x:char = 'x' // Note 4 bytes unicode

boolean

let g:bool = false; // Note 4 bytes unicode

Operators

Does not support -- and ++ but does support

a -= 2;

Remainder can be calculated using

a%3

Bitwise

let c = 1 | 2 // | OR

Shift

let two_to_10 = 1 << 10; // 1024

Logical of standard e.g.

let pi_less_4 = std::f64::consts::PI < 4.0; // true

Scope and shadowing

Curly braces keep scope

 fn test()
 {
   {
     let a = 5; 
   }
   println!("Broken {a}");
 }

Shadowing is fine though

 fn test()
 {
   let a = 5; 
   {
     let a = 10; 
     println!("10 {a}");
   }
   println!("5 {a}");
 }

Constants

Standard const

 const MEANING_OF_LIFE:u8 = 42;

Static const

 static Z:i32 = 123;

Stack and Heap

Same a c++ i.e.

 let y = Box::new(10);
 println!("y = {}", *y);

Types

Tuples

Eezy peezy lemon squeezy

fn sum_and_product(x:i32,y:i32) -> (i32, i32)
{
 (x+y, x*y)
}

fn main()
{
  let sp = sum_and_product(3,4);
  let (a,b) = sp;
  let sp2 = sum_and_product(4,5);
   
  // combine
  let combined = (sp, sp2);
  let ((c,d), (e,f)) = combined;
}

Arrays

Array sizes cannot grow in rust

Simple

let mut a:[i32;5] = [1,2,3,4,5];
// Or 
let mut a = [1,2,3,4,5];
// Length
 a.len()
// Assignment
 a[0] = 321
// Printing
 println!("{:?}", )
// Testing
  if a == [1,2,3,4,5]
  {
  }
// Initialise
  let b = [1,10]; // 10 array initialised to 1

Multi Dimension

Here is a two dimension array

let mtx:[[f32;3];2] =
[
  [1.0, 0.0, 0.0],
  [0.0, 2.0, 0.0],
];

Slices

A slice is a non-owning pointer to a block of memory. For example

// Create a vector
let v: Vec<i32> = {0..5}.collect();

// Now create a slice (reference)
let sv: &[i32]= &v;

// We create a slice with only some elements
let sv1: &[i32]= &v[2..4];

// Printing these will produce the same result
println!("{:?}",v);
println!("{:?}",sv);

// And the range
println!("{:?}",sv1);

Get the first 3 elements of an array

fn use_slice(slice: &mut[i32])
{

}

fn test()
{
  let mut data = [1,2,3,4,5];
  // Passes element 1-3 to use_slice as a reference
  use_slice( &mut data[1..4]); 
}

Strings

Basic String

let name = String::from("Iain");

Two types, static string and string type

let s = "hello";
// Cannot do
// let h = s[0]
// You can iterate as a sequence using chars e.g.
for c in s.chars()
{
  println!("{}", c);
}

And now the mutable string in rust essentially an vector // Create a string

let mut letters = String::new();

Add a char

let a = 'a' as u8;
letters.push(a as char);

String to str

let u:%str = &letters;

Concatenation

let z = lettters + &letters

Other examples

let mut abc = "hello world".to_string()'
abc.remove(0);
abc.push_str("!!!");
abc.replace("ello","goodbye")

Hashmap

Reminds me of my C++ and Java days. No surprises here for reference

let mut basket = HashMap::new();

basket.insert(String::from("banana"), 2);
basket.insert(String::from("pear"), 2);
basket.insert(String::from("peach"), 2);

Updating was a bit more tricky than expected. This was the copilot approach

struct TeamScores {
    goals_scored: u8,
    goals_conceded: u8,
}

fn build_scores_table(results: &str) -> HashMap<&str, TeamScores> {
    // The name of the team is the key and its associated struct is the value.
    let mut scores = HashMap::new();

    for line in results.lines() {
        let mut split_iterator = line.split(',');
        // NOTE: We use `unwrap` because we didn't deal with error handling yet.
        let team_1_name = split_iterator.next().unwrap();
        let team_2_name = split_iterator.next().unwrap();
        let team_1_score: u8 = split_iterator.next().unwrap().parse().unwrap();
        let team_2_score: u8 = split_iterator.next().unwrap().parse().unwrap();

        // TODO: Populate the scores table with the extracted details.
        // Keep in mind that goals scored by team 1 will be the number of goals
        // conceded by team 2. Similarly, goals scored by team 2 will be the
        // number of goals conceded by team 1.
        let team_1 = scores.entry(team_1_name).or_insert(TeamScores::default());
        team_1.goals_scored += team_1_score;
        team_1.goals_conceded += team_2_score;

        let team_2 = scores.entry(team_2_name).or_insert(TeamScores::default());
        team_2.goals_scored += team_2_score;
        team_2.goals_conceded += team_1_score;
    }

    scores
}

The suggestion was to use get_mut on hashmap but struggle to get this to work. The solution from Chris biscardi on youtube was this, clearly the rust team looked at this and did it better.

fn build_scores_table(results: &str) -> HashMap<&str, TeamScores> {
...
        scores
            .entry(team_1_name)
            .and_modify(|team: &mut TeamScores| {
                team.goals_scored += team_1_score;
                team.goals_conceded += team_2_score;
            })
            .or_insert(TeamScores {
                goals_scored: team_1_score,
                goals_conceded: team_2_score,
            });

        scores
            .entry(team_2_name)
            .and_modify(|team: &mut TeamScores| {
                team.goals_scored += team_2_score;
                team.goals_conceded += team_1_score;
            })
            .or_insert(TeamScores {
                goals_scored: team_2_score,
                goals_conceded: team_1_score,
            });

Control Flow

if statement

Same as C++ except no brackets

 if temp > 30 
 {
    println!("Blah");
 }
 else if temp < 10 
 {
    println!("Blah"); 
 }
 else
 {
    println!("Blah"); 
 }

Elvis is like

  let a = if temp > 30 {"sunny"} else {"cloud"}

While and Loop

While

Same as C++ except no brackets

 while x < 1000
 {
 }

There is support for continue and break

Loop

Loop is while true

 loop
 {
    if y == 1 << 10 { break; }
 }

For Loop

A bit like kotlin loops (I think)

 for x in 1..11
 {
    println!("x = {}",x);
 }

You can get position in series as well

 for (pos,x) in (1..11).enumerate()
 {
    println!("x = {}, pos = {}",x, pos);
 }

Rust Principles

Move

Move is when you assign a value to another variable. If we try and use a variable after the move we will get an error.

let v = vec![1,2,3]
let v2 = v;
println!("{:?}",v2)
println!("{:?}",v) // Error

Copy

When we copy something me make a new thing. They is not the same a let a = b, which is assignment. Copy means we duplicate the underlying data of the type. For primitives a copy is implemented by default. This is because the primitive has a know size. E.g. u32, bool etc. If you want to be able to copy a non primitive you need to add the derive macro. Note Clone must also be specified

#[derive(Copy, Clone)]
enum Direction {
    North,
    East,
    South,
    West,
}

#[derive(Copy, Clone)]
struct RoadPoint {
    direction: Direction,
    index: i32,
}

Clone

Clone is a method you can call on a struct if you want a second instance and not move the ownership. Here is an example. The struct obviously needs to implement the Copy/Clone macro. Cloning clearly increases the memory used.

let v = vec![1,2,3]
let v2 = v.cone();
println!("{:?}",v)
println!("{:?}",v2)

References

So references are like C++ references, but for rust this means you can pass the ownership during function call

main() {
    let mut s = String::from("Hello");
    change_string(&mut s);
}

fn change_string(some_string: &mut String) {
    some_string.push_str(", world!");
}

Note for returning a Reference
If we are returning a reference we must be returning a parameter as all local variables are destroyed. (Clearly Rust is not going to allow new MyMemory(6502)

Structs

General

There are 3 types of structs, name, tuple and unit structs

• • Named
  • Tuples
  • Unit

Name Struct

struct User
{
  active: bool,
  username: String,
  sign_in_count: u32
}

let user1 = User{active: true, username: String::from("Biil"),
  sign_in_count: 0};
println!("{}", user1.username);

...

fn build_user(username: String) -> User {
   User {
     username,
     active:true,
     sign_in_count: 1
   }
}

Tuple Struct

Tuple structs use the order in which declared to assign.

   struct Coordinates{i32,i32,i32};
   let coords = Coordinates{1,2,3};

Unit Struct

These are used to mark the existence of something

struct UnitStruct;
let a = UnitStruct{}

The example shown was when you are implementing a trait (interface) but the properties were not required for this type. So given a trait for Area, Square uses size but Point does not have an area as it is zero

trait AreaCalculator {
  fn calc_area(&self) => f64
}

struct Square {
  size: f64
}

struct Point;

impl AreaCalculator for Square { 
  fn calc_area(&self) -> f64 {
    self.size * self.size
  } 
} 

impl AreaCalculator for Point { 
  fn calc_area(&self) -> f64 {
    0.0
  } 
}

We can use it for error

struct DivideByZero;

fn divide(nom: f64, den: f64) -> Result<f64, DivideByZero> {
   if den != 0.0 {
       Ok(nom/den)
   } else {
       Err(DivideByZero)
   }
}

Example Structs

struct Point
{
  x: f64,
  y: f64
}

fn main()
{
  let p = Point { x: 30.0, y: 4.0 };
  println!("point is at ({},{})", p.x, p.y)
}

Methods on Structs

Methods on struct require the first argument to be self

Example Method

Add method len to struct

struct Line
{
 start: Point,
 end: Point
}

// Declare impl using the keyword impl. Not ends with no semi colon.
impl Line
{
  fn len(&self) -> f64
  {
    let dx = self.start.x - self.end.x;
    let dy = self.start.y - self.end.y;
    (dx*dx+dy*dy).sqrt()
  }
}

Changing an attribute

To change an attribute and ensure you do not break the borrowing rules we do

struct Square {
   width: u32,
   height: u32
}

impl Square
{
  fn area(&self) -> u32 {
     self.width * self.height
  }

  fn change_width(&mut self, new_width: u32) -> Self
  {
    self.width = new_width;
  }
}

...
main() {
...
   let mut sq = Square(width:5, height: 5);
   sq.change_point(10) 
}

Lifetime

What are Dangling References

The code below will not compile. This is because x goes out of scope before r. I am guessing this is what is known as a dangling reference.

fn test() {
    let r;
    {
        let x = 5;
        r = &x; // Error `x` does not live long enough
    }
    log::info!("{}",r);
}

Lifetime Annotations

Not sure which way around these are but you specify lifetime annotations on functions and structs and they imply information to the compiler on how long the parameters will live for.

Three Rules of Lifetimes

Here are the rules but we also need to understand what they apply to. Kind of chicken and egg. An example is give below which is broken because these rules are not followed.

1. Each Parameter that is a reference gets its own lifetime parameter
2. If there is exactly one input lifetime parameter, that lifetime is assigned to all output lifetime parameters
3. If there are multiple input lifetime parameters, but one of them is &self or &mut self the lifetime is assigned to all output lifetime parameters

Example (Broken code)

Here is an example of code which cannot be compiled without lifetime being specified.

pub struct TestStruct {
    length: i32,
}

fn test2(x: &TestStruct, y: &TestStruct) -> &TestStruct { // Missing lifetime specifier
    if x.length > y.length {
      x
    }
    else {
      y 
    }
}

Adding Annotations

To do this we specify annotations. The extension in vscode does this for us using the quick fix. The code now looks like this

fn test2<'a>(x: &'a TestStruct, y: &'a TestStruct) -> &'a TestStruct {
    if x.length > y.length {
      x
    }
    else {
      y 
    }
}

My inference from this is that all parameters have the same lifetime.

Lifetime Annotations for Structs

Structs can also have lifetime annotations. If you specify a reference then you will need to specify a lifetime annotation. In the example below when we make the struct of type MyString we need to make sure that str1 does not go out of scope while x of type MyString exists otherwise it would refer to something no longer in scope.

// Without lifetime annotation will not compile.
// struct MyString {
//   text: &str,
// }


struct MyString<'a> {
  text: &'a str,
}

fn main() {

    let str1 = String::from("This is my String);
    let x = MyString(text: str1.as_str());
}

Static Lifetimes

We can also have lifetimes for statics.

let s: &'static str = "I live forever";

Doing this means the values are stored in the binary.

Enums

Example 1 with Method

Seems a bit C++ but...

enum Pet {dog, cat, fish}

And now lets add a method as we do with structs. Note for this method we are returning something and with rust all locals are destroyed on return so we need to specify a lifetime.

enum Pet {dog, cat, fish}

impl Pet {
   fn what_am_i(self) -> &'static str {
      match self {
         Pet::dog => "I am a dog", 
         Pet::cat => "I am a cat",
         Pet::fish => "I am a fish",   
      }
   }
}

Example 2

enum Color {
  Red,
  Green,
  Blue
}

fn main()
{
  let c:Color = Color::Red;
  match c
  {
     Color::Red => prinln!("Color is Red");
     Color::Green => prinln!("Color is Green");
  }
}

Example 3 with Types

enum Color {
  Red,
  Green,
  Blue,
  RgbColor(u8,u8,u8) // Tuple
  CmykColor{cyan:u8, magenta:u8, yellow:u8, black:u8,} // Struct
}

fn main()
{
  let c:Color = Color::RgbColor(10,0.0);
  match c
  {
     Color::Red => prinln!("Color is Red");
     Color::Green => prinln!("Color is Green");
     Color::RgbColor(0,0,0) => prinln!("Color is Black");
     Color::RgbColor(r,g,b) => prinln!("Color is {},{},{}", r,g,b);
  }

  let d:Color = Color::CmykColor(cyan:0, magenta:0, yellow:0, black:0);
  match d
  {
     Color::Red => prinln!("Color is Red");
     Color::Green => prinln!("Color is Green");
     Color::RgbColor(0,0,0) => prinln!("Color is Black");
     Color::CmykColor(cyan:_, magenta:_, yellow:_, black:255) => prinln!("Black");
  }
}

Option<T> Enum

This enum if provided for us by rust and looks like this

enum Option<T> {
    None,
    Some(T)
}

We would choose this type when we have a case where there could be a value or not. I guess this is the equivalent of string? in Typescript where we may or may not have a value. In rust we use match to support this type.

  let some_number = Some(5); 
  let some_string = Some("a string");
  let nothing: Option<i32> = None;

Pattern Matching

Match is Exhaustive approach to pattern matching. I.E. you need to specify something for every option you are using match for. However you can include a default. I find this a great approach

Examples

Simple Match

match x
{
  0 => "zero"
  1 | 2 => "one or two"
  9...11 => "lots of"  // two dots does not include end value (exclusive)
  _ if(blahh) => "something"
  _ => "all others"
}

Here is another example.

 let country = match country_code
 {
    44 => "uk",
    46 => "sweden",
    7 => "russia"
    1...999 => "unknown" // other triple dot does include end value (inclusive)
    _ => "invalid" // invalid
 };

This just shows inclusive which is ..= unlike kotlin which I think is 3 dots

// This function returns how much icecream there is left in the fridge.
// If it's before 22:00 (24-hour system), then 5 scoops are left. At 22:00,
// someone eats it all, so no icecream is left (value 0). Return `None` if
// `hour_of_day` is higher than 23.
fn maybe_icecream(hour_of_day: u16) -> Option<u16> {
    match hour_of_day {
        0..22 => Some(5),
        22..=23 => Some(0),
        _ => None,
    }
}

More Complex

Stumped me when see thing for the first time prior to type script and possibly lambda. Here we define anonymous functions which match the type of the enum. Here is the enum which is used in another struct

enum Message {
    Move(Point),
    Echo(String),
    ChangeColor(u8, u8, u8),
    Quit,
    Resize { width: u64, height: u64 },
}

It has functions for each enum type.

struct State {
    width: u64,
    height: u64,
    position: Point,
    message: String,
    // RGB color composed of red, green and blue.
    color: (u8, u8, u8),
    quit: bool,
}

impl State {
    fn resize(&mut self, width: u64, height: u64) {
        self.width = width;
        self.height = height;
    }

    fn move_position(&mut self, point: Point) {
        self.position = point;
    }

    fn echo(&mut self, s: String) {
        self.message = s;
    }

    fn change_color(&mut self, red: u8, green: u8, blue: u8) {
        self.color = (red, green, blue);
    }

    fn quit(&mut self) {
        self.quit = true;
    }

    fn process(&mut self, message: Message) {
...
    }
}

At first I struggled to understand how to implement process but all you need to do is provide an ()_=> {} for each type. For Quit I completely understood but for the others was confused. Obvious once you know and I am sure copilot will do this for me

    fn process(&mut self, message: Message) {
        match message {
            Message::Move(point) => self.move_position(point),
            Message::Echo(output) => self.echo(output),
            Message::ChangeColor(red, green, blue) => self.change_color(red, green, blue),
            Message::Quit => self.quit(),
            Message::Resize { width, height } => self.resize(width, height),
        }
    }

Match on Tuples

This is an exert from [Game of Life]. We can match on tuples, and I imagine other types too. For tuples you can specify a value or compare to a value. Note the use of otherwise

    let next_cell = match (cell, live_neighbors) {
      // Rule 1: Any live cell with fewer than two live neighbours
      // dies, as if caused by underpopulation.
      (Cell::Alive, x) if x < 2 => Cell::Dead,
      // Rule 2: Any live cell with two or three live neighbours
      // lives on to the next generation.
      (Cell::Alive, 2) | (Cell::Alive, 3) => Cell::Alive,
      // Rule 3: Any live cell with more than three live
      // neighbours dies, as if by overpopulation.
      (Cell::Alive, x) if x > 3 => Cell::Dead,
      // Rule 4: Any dead cell with exactly three live neighbours
      // becomes a live cell, as if by reproduction.
      (Cell::Dead, 3) => Cell::Alive,
      // All other cells remain in the same state.
      (otherwise, _) => otherwise,
};

Operators and Symbols

Found in Table B-1 here [Operators and Symbols]

1. [Range]: 1..10
2. [RangeFrom]: 1..
3. [RangeTo]: ..10
4. RangeFull: ..
5. RangeInclusive: 1..=10
6. RangeToInclusive: ..=10

Option <T> and if let

Used to avoid null or invalid values. This was used in things where the value might be present. Maybe command line arguments where some were provide or none were provided. Lets to the classic divide by zero.

let x = 3.0
let y = 0.0 // Divide by zero

let result:Option<f64> = 
   if y != 0.0 { Some(x/y) } else { None };

// Using match
match result {
   Some(z) => println!("Goody result"),
   None => println!("No result")
}

// Using if let
if let Some(z) = result { println!("z = {}", z); }

More if let

Here is another example

let mut stack = Vec:new();
stack.push(1);
stack.push(2);
stack.push(3);
while let Some(top) = stack.pop() {
   println!("{}", top);
}

while let

The above example makes great sense but while doing rustlings the was this question

  // TODO: Make this a while-let statement. Remember that `Vec::pop()`
  // adds another layer of `Option`. You can do nested pattern matching
  // in if-let and while-let statements.
  integer = optional_integers.pop() {
    assert_eq!(integer, cursor);
    cursor -= 1;
  }

I did like the Some Some approach

   while let Some(Some(integer)) = optional_integers.pop() {
     assert_eq!(integer, cursor);
     cursor -= 1;
   }

But could not get the You can do nested pattern matching in if-let and while-let statements to look nice

   while let Some(integer) = if let Some(integer) = optional_integers.pop() {
     integer
   } else {
     None
   } {
     assert_eq!(integer, cursor);
     cursor -= 1;
   }

Just wouldn't let it lie, found a way to turn it up the right way

   while let Some(integer) = optional_integers.pop() {
     if let Some(integer) = integer {
       assert_eq!(integer, cursor);
       cursor -= 1;
     }
   }

Generics

Simple

This is very similar to C++ Templates and TypeScript Generics

struct Point<T>
{
  x: T,
  y: T
}

fn generics()
{
  let a:Point<i32> = Point {x: 0, y: 4}
}

Using Implementation

Must the same, just need good examples and we a well away

struct Wrapper<T> {
    value: T,
}

impl<T> Wrapper<T> {
    fn new(value: T) -> Self {
        Wrapper { value }
    }
}

Traits

Traits are similar to interfaces in java and c#

Defining a Traits

trait Animal
{
  fn create(name:&'static str);

  fn name(&self) => &'static str;

  fn talk(&self)
  {
     println!("{} cannot talk",self.name()); 
  }
}

Implement a Trait

Here we create a struct which will implement out trait. Note we do not have to implement all functions if the trait provides a default implementation

Implement a Trait for Animal

struct Human
{
   name: &'static str;
}

impl Animal for Human
{
  fn create(name:&'static str) -> Human
  {
    Human{name: name}
  }

  fn name(&self) -> &'static str
  { 
     self.name
  }
  // override default
  fn talk(&self)
  {
     println!("{} can talk",self.name()); 
  }
}

Implement a Trait for Cat

Here we implement the Animal Trait for Cat

struct Cat
{
   name: &'static str;
}

// Implement interface
impl Animal for Cat
{
  fn create(name:&'static str) -> Cat
  {
    Cat{name: name}
  }

  fn name(&self) -> &'static str
  { 
     self.name
  }
  // override default
  fn talk(&self)
  {
     println!("{} says meeow",self.name()); 
  }
}


// Usage
let h:Human = Animal::create("John");
let c:Cat = Animal::create("John");

Default Trait and Spread

For a struct we can create a default for it. We can use a typescript like spread operator (although it must be last) for override these defaults

pub struct Circle {
    color: String,
    point: Point,
    radius: u16,
}

impl Circle {
    pub fn new(color: String, point: Point, radius: u16) -> Circle {
        Circle {
            color,
            point,
            radius,
        }
    }

    pub fn default_color(point: Point, radius: u16) -> Circle {
        Circle {
            point,
            radius,
            ..Default::default()
        }
    }
}

impl Default for Circle {
    fn default() -> Self {
        Circle {
            color: String::from("black"),
            point: Point::new(0, 0),
            radius: 0,
        }
    }
}

// Default Circle
let circle = Circle::default();

// Default Black Circle
let circle = Circle::default_color(Point::new(1, 1), 1);

Traits and Impl

To allow any struct which implements the trait we use the dyn keyword

trait Licensed {
    fn licensing_info(&self) -> String {
        "Default license".to_string()
    }
}

struct SomeSoftware;
struct OtherSoftware;

impl Licensed for SomeSoftware {}
impl Licensed for OtherSoftware {}

// TODO: Fix the compiler error by only changing the signature of this function.
fn compare_license_types(software1: impl Licensed, software2: impl Licensed) -> bool {
    software1.licensing_info() == software2.licensing_info()
}

// Now we can do this
compare_license_types(SomeSoftware, OtherSoftware)
compare_license_types(OtherSoftware, SomeSoftware)

Provided Traits

Drop Trait

Drop trait is called automatically to free up resources but you can write your own e.g. for the example above we could write

impl Drop for Course {
  fn drop(&mut self) {
     println("Dropping")
  }
}

Clone Trait

Like the drop trait we can implement our own. Refer to the clone trait for this.

Copy Trait

We can either specify #[derive(Copy, Clone)] or implement our own. There are restrictions on this

From and Into Trait

This allow us to convert from one type to another

fn into(self) -> T
fn from(T) ->  Self
fn try_into(self) -> Result<T, Self: Error>
fn try_from(value: T) -> Result<Self, Self: Error>

Trait Bounds 1

In order to allow use of more than on trait in a function we can use the +. This example means that item must implement both traits, i.e. SomeTrait and OtherTrait

fn some_func(item: impl SomeTrait + OtherTrait) -> bool {
    item.some_function() && item.other_function()
}

Trait Bounds 2

Here is an example of doing the same thing in two ways. Because we can have anything in grade (T) we must make an implementation for std::fmt::Display. That way if we make a ReportCard with a generic which does not support Display, it will not compile

struct ReportCard<T> {
    grade: T,
    student_name: String,
    student_age: u8,
}

// Approach 1
impl<T> ReportCard<T>
where
    T: std::fmt::Display,
{
    fn print(&self) -> String {
        format!(
            "{} ({}) - achieved a grade of {}",
            &self.student_name, &self.student_age, &self.grade,
        )
    }
}

// Approach 2
impl<T: std::fmt::Display> ReportCard<T> {
    fn print(&self) -> String {
        format!(
            "{} ({}) - achieved a grade of {}",
             &self.student_name, &self.student_age, &self.grade,
        )
    }
}

Trait Bounds

The example above has two ways to achieve the same thing. If we constrain what this allowed, this is called trait bounds. Lets add a second parameter.

// This example only forces the struct to implement the trait
// fn overview(item1: &imp Overview, item2: &imp Overview) 

// But this force the struct to be of the same type
// fn overview<T: Overview>(item1: &T, item2: &T)

We can add more constraints with the + operator. Now they need the second trait.

// fn overview(item1: &imp Overview + AnotherTrait, item2: &imp Overview + AnotherTrait) 
// fn overview<T: Overview + AnotherTrait>(item1: &T, item2: &T)

Here we have an example of ensuring that the incoming parameters are constrained to be of type T

struct Pointy<T> {
    x: T,
    y: T,
}

impl <T> Add for Pointy <T>
where T: Add<Output = T>
{
    type Output = Self;

    fn add(self, other: Self) -> Self {
        Self {
            x: self.x + other.x, 
            y: self.y + other.y,
        }
    }
}

Passing Trait as Parameters

So here is an example of two structs with overview implement,one using the trait default implementation, the other its own. We can use the trait similar to a pointer to a function.

Example

trait Overview {
  fn overview(&self) -> String {
     format("This is a rust course")
  } 
}

struct Course {
  headline: String,
  author: String
}

struct AnotherCourse {
  headline: String,
  author: String
}

impl Overview for Course {
}

impl Overview for AnotherCourse {
  fn overview(&self) -> String {
     format("{}, {}", self.author, self.headline)
  } 
}

We can use the overview trait a a fn parameter with

fn call_overview(item: &imp) {
   println("Overview: {}", item.overview())
}

// OR 
fn call_overview<T: Overview>(item: &T) {
   println("Overview: {}", item.overview())
}

Passing Traits (From Youtube)

Taken from Youtube and repeated. There are two notations for passing a trait. These are the same but the first is perhaps more readable. The second is known as a trait bound.

pub fn foo (traitor: &impl SpiDevice) {

}

pub fn foo<T: SpiDevice>(traitor: &T) {

}

With the impl syntax we can make the parameter have more the one trait with a plus.

pub fn foo (traitor: &impl SpiDevice + AnotherTrait) {

}

With the second syntax if we have two parameters if allows us to make sure they both share the same trait easily as the type is only specified once

pub fn foo<T: SpiDevice>(traitor1: &T, traitor2) {

}

We can also add a second trait with this syntax too.

pub fn foo<T: SpiDevice + AnotherTrait>(traitor1: &T, traitor2) {

}

This starts to get messy to we can tidy this up with the Where Clause

pub fn foo<T, U>(traitor1: &T, traitor2: &U) -> i32
where 
    T: SpiDevice + AnotherTrait,
    U: AnotherTrait + YetAnotherTrait
        {
            42
        }

Returning Traits (From Youtube)

We can also return traits but you cannot return different types which share the same trait at this time.

pub fn foo() -> SpiDevice {
    // Must be of same type
}

Common Collections

Vectors

Same a c++

let mut a = Vec::new()
a.push(1);
a.push(2);
a.push(3);
// Print
println!("a[0] {}", a[0]);

// We can create vector with initial capacity
let mut b = Vec::<i32>::with_capacity(2);

// We can initialize using an iterator values of 0-4
let c: Vect<i32> = (0..5).collect();

// Using get returns a option
match a.get(3333)
{
...
}

// Removing, pop returns an option   
let last_elem = a.pop();

// Using the option type iterating over vector to print it
while let Some(x) = a.pop()
{
   println!("x = {}",x);
}

Binary Heap

This make sure the highest is at the top. It has a peek function to allow you to peek at values.

let mut bHeap = BinaryHeap::new();
bHeap.push(1);
bHeap.push(18);
bHeap.push(20);
bHeap.push(5);
bHeap.pop();

println!("{:?}", bHeap); // 20

Maps

Not discussed

Sets

Not discussed

Error Handling

Panic

Panic happens when unhandled error occurs. This happens for instance when we access out of bounds array. We can get a backtrace by setting the environment export RUST_BACKTRACE=-1

Result Enum

The Result an enum which has two generics Result<T, E> where T is the type an E is the error. In rust we use the match to determine what to do.

let file = File::Open("Does_not_exist.mp3");
let file match file {
    Ok(file) => file,
    Err(error) => panic("Error: {:?}", error),
};

Mapping Errors

Rust likes you to make your own errors and map the ones you handle to you errors which makes sense. We make our own errors using enums

enum ParsePosNonzeroError {
    Creation(CreationError),
    ParseInt(ParseIntError),
}

Now we can provide helper function to convert from one type of error to ours

impl ParsePosNonzeroError {
    fn from_creation(err: CreationError) -> Self {
        Self::Creation(err)
    }

    // TODO: Add another error conversion function here.
    fn from_parse_int(err: ParseIntError) -> Self {
        Self::ParseInt(err)
    }
}

Now in our parse function we can map the errors in parse() to our own

#[derive(PartialEq, Debug)]
struct PositiveNonzeroInteger(u64);

impl PositiveNonzeroInteger {
    fn new(value: i64) -> Result<Self, CreationError> {
        match value {
            x if x < 0 => Err(CreationError::Negative),
            0 => Err(CreationError::Zero),
            x => Ok(Self(x as u64)),
        }
    }

    fn parse(s: &str) -> Result<Self, ParsePosNonzeroError> {
        // TODO: change this to return an appropriate error instead of panicking
        // when `parse()` returns an error.
        let x: i64 = s.parse().map_err(ParsePosNonzeroError::from_parse_int)?;
        Self::new(x).map_err(ParsePosNonzeroError::from_creation)
    }
}

Testing

We specify the cfg option and use the assert library

fn sqrt(number: f64) -> Result<f64, String> {
    if number >= 0.0 {
        Ok(number.powf(0.5))
    } else {
        Err("negative floats don't have square roots".to_owned())
    }
}

#[cfg(test)]
mod tests {
    use super::*;

    #[test]
    fn test_sqrt() -> Result<(), String> {
        let x = 4.0;
        assert_eq!(sqrt(x)?.powf(2.0), x);
        Ok(())
    }
}

We can run these with

cargo test

Handling CLI Arguments

Clap seems to be the approach for this, make sure you install the macros otherwise errors will show on the derive macro.

cargo add clap --features derive

And the code. The shows how to replace text from a file to another file.

use regex::Regex;
use text_colorizer::*;
use clap::Parser;
use std::fs;

#[derive(Parser,Debug)]
struct Args {
    #[clap(short, long)]
    pattern: String,
    #[clap(short, long)]
    replace: String,
    #[clap(short, long)]
    input_file: String,
    #[clap(short, long)]
    output_file: String,
}

fn replace (pattern: &str, replace: &str, data: &str) -> Result<String, regex::Error> {
    let regex = Regex::new(pattern)?;
    Ok(regex.replace_all(data, replace).to_string())
}

fn read_write_file(args: Args) {

    let data = match fs::read_to_string(&args.input_file) {
        Ok(data) => data,
        Err(err) => {
            eprintln!("{} failed to read from file {}: {:?}", 
                "Error".red().bold(), 
                args.input_file, err);
            std::process::exit(1);
        }
    };

    let replace_data = match replace(&args.pattern, &args.replace, &data) {
        Ok(data) => data,
        Err(err) => {
            eprintln!("{} failed to replace text {:?}", 
                "Error".red().bold(), 
                err);
            std::process::exit(1);
        }
    };

    match fs::write(&args.output_file, replace_data) {
        Ok(_) => {},
        Err(err) => {
            eprintln!("{} failed to write to file {}: {:?}", 
                "Error".red().bold(), 
                args.output_file, err);
            std::process::exit(1);
        }
    };
}

fn main_body(args: Args) -> Result<(), ()> {
    // Your main body here
    println!("input_file {}", args.input_file);
    println!("output_file {}", args.output_file);
    read_write_file(args);
    Ok(())
}

fn main() {
    let args = Args::parse();
    main_body(args).unwrap_or_else(|_| {
        eprintln!("Error");
        std::process::exit(1);
    });
}

Test Section Working 1

Test Text 1

enum ParsePosNonzeroError {
    Creation(CreationError),
    ParseInt(ParseIntError),
}

Test Text 2

impl ParsePosNonzeroError {
    fn from_creation(err: CreationError) -> Self {
        Self::Creation(err)
    }
}

Test Text 3

#[derive(PartialEq, Debug)]
struct PositiveNonzeroInteger(u64);

impl PositiveNonzeroInteger {
    fn new(value: i64) -> Result<Self, CreationError> {
        match value {
            x if x < 0 => Err(CreationError::Negative),
            0 => Err(CreationError::Zero),
            x => Ok(Self(x as u64)),
        }
    }
}

Closures

Closures are functions you which you can use the available scope to with that function. They look like anonymous function in typescript

Simple Example of a closure

The adder closure takes a parameter and captures existing scope at creation to produce the answer 17.

    let x: i32 = 5;
    let adder = |a| a + x;
    let b = adder(12);
    print!("b {}", b)

Traits provided for closures

These traits are provided in rust

• Fn
• FnOnce
• FnMut

We need to provide some explanation for this

/*
  || drop(v)             FnOnce as we only drop once
  |args| v.contains(arg) Fn as does not modify
  |args| v.push(arg)     FnMut as it does modify 
*/

Mapper Function using Closure

Here is a closure which is like typescript map

things.map((element: number) => element * 2)

And in rust

things.iter().map(|element| element * 2).collect()

DELETE Simple Closure

test0

    let x: i32 = 5;

test1

    let x: i32 = 5;

test2

    let x: i32 = 5;
    let b = adder(12);
    let x: i32 = 5;
    let adder = |a| a + x;
    let b = adder(12);
    print!("b {}", b)

Original Code

test

    let x: i32 = 5;
    let adder = |a| a + x;
    let b = adder(12);
    print!("b {}", b)

More Complex

We use the existing struct, make a copy and call them inside the to pass in objects from somewhere and use their results with the closure function

fn create_percent_complete_fn<'a>(
  &'a self,
  looper: &'a LooperDuration,
  interpolator: &'a AccelerateDecelerateInterpolator,
) -> impl Fn(u16) -> f64 + 'a {
  let looper_ref = looper;
  let interpolator_ref = interpolator;
  let closure = move |elapsed_time: u16| {
    let rate_of_change = looper_ref.get_t_with_elapsed_time(elapsed_time);
    let b = interpolator_ref.get_interpolar(rate_of_change);
    return b * 100 as f64;
  };
  return closure;
}

Iterators

Nothing fancy here except we have to make the iterator mut (not surprising but worth mentioning). An iterator is any type which implements the iter trait and an iterable is a type that implements into iterator.

let vec = [1, 2, 3];
let mut iter = vec.iter();
while let Some(i) = iter.next() {
    println!("i {}", i);
}

Just like Typescript we can now use closures like the map, filter functions in typescript. E.g. given a vector of times we can filter on property

items.into_iter().filter(|i|i.name == search_name).collect()

We can implement our own iterator on our struct like below

struct Range {
    start: u32,
    end: u32
 }


impl Iterator for Range {
    type Item = u32;
    fn next(&mut self) -> Option<Self::Item> {
       if self.start >= self.end {
          return None 
       }  
       let result = Some(self.start);
       self.start += 1;
       result
    }
 }

Example 1 Iterators

This iterates over the input slice, calling next() once. At this point first contains the first item in input and chars now contains the rest. We concatenate the two values.

fn capitalize_first(input: &str) -> String {
    let mut chars = input.chars();
    match chars.next() {
        None => String::new(),
        Some(first) => {
            let capital = first.to_uppercase();
            format!("{}{}", capital, chars.as_str())
        }
    }
}

Example 2 Iterators Mapping over slice

This is a good example of mapping like typescript but over a slice

fn capitalize_words_vector(words: &[&str]) -> Vec<String> {
    words
        .iter()
        .map(|&element| capitalize_first(element))
        .collect::<Vec<_>>()
}

Pointers

This is the same a C++ with some rusty new language. The Box is a pointer type that uniquely owns a heap allocation of type

Box Type

This the basic pointer

let t = (12, "eggs");
let b = Box::new(t);
println!("b {:?}", b); // (12, "eggs")

RC Type

This allows multiple pointers to the same thing in memory. For example

let s1 = Rc::new(String::from("Pointer));
let s2 = s1.clone();
let s3 = s2.clone();

println("{},{},{}",s1,s2,s3) // Pointer, Pointer, Pointer

RefCell

A RefCell is another way to change values without needing to declare mut. It means "reference cell", and is like a Cell but uses references instead of copies, Rust refers to Interior mutability which seems to say it is a pattern where the user is allow to modify data despite the data not having a mut associated with it. The function of RefCell allow us to modify when we dereferenced.

use std::{cell::RefCell};

struct Flagger {
    is_true: RefCell<bool>,
 }

// borrow returns Ref<T>
// borrow_mut returns RefMut<t>

let flag = flagger{ is_true: RefCell::new(true)};

// let reference =  flag.is_true.borrow();
// println!("{}", reference);

let mut mut_ref = flag.is_true.borrow_mut();
*mut_ref = false;

Note if we want to use it twice, i.e. uncomment the println we have to wrap the RefCell with RC

   // In struct
   is_true: Rc<RefCell<bool>>

   // Initialize
   let flag = flagger{ is_true: Rc::new<RefCell::new(true))};

Concurrency

Thread Join

Threads are similar to C++ and C#. Let do a basic join

let handle = thread::spawn(move || {
    println!("Hello from a thread!")
});

handle.join().unwrap();
println!("Hello Main!")

Not threads do not always finish in the order created

    v.into_iter().for_each(|e| {
        thread_handles.push(thread::spawn(move || println!("Thread {}",e)));
    });

    thread_handles.into_iter().for_each(|handle| {
        handle.join().unwrap();
    });

// Thread 2
// Thread 1
// Thread 3

Channels

This is like channels in kotlin. These live in the mpsc namespace which stands for multi producer single consumer. The value being sent is taken on send and receive. I.E. you cannot use the value being sent after send.

Example One Producer

use std::sync::mpsc::channel;
use std::thread;

let (sender, receiver) = channel();

// Spawn off an expensive computation
thread::spawn(move|| {
    sender.send(expensive_computation()).unwrap();
});

// Do some useful work for awhile

// Let's see what that answer was
println!("{:?}", receiver.recv().unwrap());

Example Two Producer

We can have multiple producers and one receiver. To ensure that the receiver is not overwhelmed rust provides a sync_channel method on the mpsc. When used the sender will block when the queue if full and automatically continue when reduced.

    let (producer1, receiver) = sync_channel(1000);
    let producer2 = producer1.clone();

    // Send from Producer 1
    thread::spawn(move|| {
        let vec = vec![String::from("transmitting"), String::from("hello"), String::from("world"),];
        for val in vec {
            producer1.send(val).unwrap();
        }
    });

    thread::spawn(move|| {
        let vec = vec![String::from("producer2"), String::from("hello"), String::from("world 2"),];
        for val in vec {
            producer2.send(val).unwrap();
        }
    });

    // Send from Producer 
    for received in receiver {
        println!("Got: {}", received);
    }

Sync and Send Type Traits

Send

Types that implement send are safe to pass by value to another thread and moved across threads. Almost all types implement send but there are exceptions, e.g. Rc however the Atomic Reference Counter (Arc) can be. This did not compile for me if I tried to use Rc with the error std::rc::Rc<std::string::String>` cannot be sent between threads safely within `{closure@src/main.rs:116:24: 116:31}`, the trait `std::marker::Send` is not implemented for `std::rc::Rc<std::string::String> but it did on the course.

Sync

Types that implement sync are safe to pass by non mutable reference to another thread. These types can be shared across threads.

Mutexes and Threads (Shared State)

Introduction

These a like c++ mutexes. You need to get and release a lock when using the resource. It was stressed the the atomic reference counter Arc (no Rc) is used for sharing resources across threads. Mutexes are used for mutating (modifying) data that is shared across threads.

Deadlocks

Below is an example of a deadlock where the same thread asks for the lock without releasing. Note you can release the mutex with drop on the guard (numMutexGuard). I also noticed if you did not use the value then the code did complete - maybe an optimizer.

    let counter = Arc::new(Mutex::new(0));
    let mut handles = vec![];
    for _ in 0..10 {
        let counter = Arc::clone(&counter);
        let handle = thread::spawn(move|| {
            let mut numMutexGuard = counter.lock().unwrap();
            // *numMutexGuard += 1;
            // std::mem::drop(numMutexGuard);
            let mut numMutexGuard2 = counter.lock().unwrap();
            *numMutexGuard2 += 1;
        });
        handles.push(handle);
    }

    for handle in handles {
        handle.join().unwrap();
    }

    println!("Result: {}", *counter.lock().unwrap());

Poisoned Mutex

This is a term which means when a thread is executing and has a lock but panics. The mutex is hanging or poisoned. In rust we can actually recover from this though I suspect this is very undesirable. Here is the sample code where we match on poisoned.

    let lock = Arc::new(Mutex::new(0));
    let lock2 = Arc::clone(&lock);

    let _ = thread::spawn(move || {
        let _guard = lock2.lock().unwrap(); // acquire lock
        panic!("thread1"); // mutex is not poisoned
    }).join();

    let mut guard = match lock.lock() {
        Ok(guard) => guard,
        Err(poisoned) => poisoned.into_inner(),
    };

    *guard += 1;
    println!("lock value: {}", *guard);

Rayon

Quick example of rayon for parallelization with Rayon. Was a little unhappy with the changing of signatures for reduce in rayon but ho-hum. This took 6550ms single threaded and 186ms multi-threaded.

fn factoral (n: u32) -> BigUint {
    if n == 0 || n ==1 {
        BigUint::from(1u32)
    } else {
        // Reduce in Typescript is array.reduce((acc, next_value) => acc * next_value, 1)
        (1..=n).map(BigUint::from).reduce(|acc, next_value| acc * next_value).unwrap()
        // new way 
        // n * factoral(n - 1)
    }

}

fn factoral_with_rayon (n: u32) -> BigUint {
    if n == 0 || n ==1 {
        BigUint::from(1u32)
    } else {
        (1..=n).into_par_iter().map(BigUint::from).reduce(|| BigUint::from(1u32), |acc, x| acc * x)
    }
}


main() {
    let mut now = std::time::Instant::now();
    factoral(80000);
    println!("factoral took {} seconds", now.elapsed().as_millis());
    now = std::time::Instant::now();
    factoral_with_rayon(80000);
    println!("factoral with rayon took {} seconds", now.elapsed().as_millis());
}

Asyncronous

Async and Await (Future Trait)

This is just like promises. I learned a few things doing this. Renaming namespaces can be done using as. For the asynchronous work with thread still present there was a lot of crossover between the two. Note not all code listed so additional use statements demonstrate the renaming requirement.

use async_std::{prelude::*, io as async_io, task as async_task, fs as async_fs, fs::File as async_file};
use std::{cell::RefCell, fs, thread, sync::{mpsc::sync_channel, Arc, Mutex}};

async fn read_file(path: &str) -> async_io::Result<String> {
    let mut file: async_file = async_fs::File::open(path).await?;
    let mut contents = String::new();
    file.read_to_string(&mut contents).await?;
    Ok(contents)
}

fn main() {
    let task = async_task::spawn(async {
        let  result = read_file("Cargo.toml").await;
        match result {
            Ok(k) => println!("contents {}", k),
            Err(err) => println!("error {}", err),
            
        }
    });

    async_std::task::block_on(task);
    println!("Task stopped");
}

Big O Notation

General

Big O notation is a mathematical notation that describes the limiting behavior of a function when the argument tends towards a particular value or infinity. This graph tries to convey how the performance of an algorithm is impacted based on the O(n) value. Other references are Big_o_notation and [here]
They talk about approaches used.

• Experimental
• Theoretical

Experimental

Experimental analysis involves running a program and measuring how long it takes to complete. The problem with this approach is that programs run non-deterministically and at different speeds depending on the hardware, compiler, programming language, and other factors.

Theoretical

The alternative is theoretical analysis. One approach to theoretical analysis is to approximate the running time by counting the number of steps an algorithm takes for a given input.

Recursion and Factorial

General

Recursion is when we use a function multiple times but change the input. An example of recursion is factoral where

   n! = n * n-1 * n -2 * n-2...
   n! = n * (n-1)!

when doing recursion we must specify a any base cases and a termination condition. For factorial this would be line 12

#include<stdio.h>
long int multiplyNumbers(int n);
int main() {
    int n;
    printf("Enter a positive integer: ");
    scanf("%d",&n);
    printf("Factorial of %d = %ld", n, multiplyNumbers(n));
    return 0;
}

long int multiplyNumbers(int n) {
    if (n>=1)
        return n*multiplyNumbers(n-1);
    else
        return 1;
}

Call Stack

When you use recursion we need to be careful that we do not have a stackoverflow.

Base Cases

Base Case is defined as the condition in Recursive Function, which tells the function when to stop. It is the most important part of every Recursion, because if we fail to include this condition it will result in INFINITE RECURSIONS.

Fibonacci

Here is the Fibonacci example which seems to be the darling of all CS students. Here there are 2 bases cases

• base case when n = 0
• base case when n = 1

I think the match operator is a great way to express recursive functions

fn fibonacci(n: u32) -> u32 {
    match n {
        0 => 1,
        1 => 1,
        _ => fibonacci(n - 1) + fibonacci(n - 2),
    }
}

fn main() {
    println!("Fibonacci generator");
    println!("{}", fibonacci(1));
    println!("{}", fibonacci(3));
    println!("{}", fibonacci(5));
}

Palindrome

Here we decide what the parameters are for the base cases

fn is_palindrome(s: &str, start: usize, end: usize) -> bool {
    if start >= end {
        return true;
    }

    if s.chars().nth(start) != s.chars().nth(end) {
        return false;
    }

    is_palindrome(s, start + 1, end - 1)
}

fn main() {
    let answer1 = is_palindrome("racecar", 0, 6);
    println!("{}", answer1);
    let answer2 = is_palindrome("xracecar", 0, 7);
    println!("{}", answer2);
}

Tower of Hanoi

This was a more challenging puzzle to demonstrate recursive functions. Here the object of the game is to move the blocks from tower one to tower three and replicate the order. The rules are

• Only one block at a time
• Larger blocks cannot be placed on smaller blocks

I used the youtube video [here] which provided great graphics on how to solve this. Here is a function for 3 blocks and the steps required to achieve it.

Took me a while to understand but the steps (not the moves were)

• Move n-1 Discs from A to B using C
• Move a Disc from A to C
• Move n-1 Discs from B to C using A

So in rust we can see the number of moves based on number of discs (n) is

fn toh(n: i32) -> i32 {
   if n = 0 {
      return 0
   }
          
   return toh(n-1) // Step 1 n-1 Discs)      
          + 1      // Step 2 1 disc
          toh(n-1) // Step 3 n-1 Discs)      
}

fn main() {
    println!("{}", toh(4)); // 15
}

Sum of Triangles

This is where we start with the base and the next row is the some of it child numbers. All my own work this time.

fn sum_of_triangles(numbers: &mut Vec<i32>, current_sum: i32) -> i32 {

    println!("Input Current Sum {:?}", current_sum);

    // Empty return
    if numbers.len() == 1 {
        return current_sum;
    }

    // Create a new vector
    let mut new_vector = Vec::<i32>::new();

    // Iterate through the vector - 1
    for n in 0..numbers.len() -1 {
        new_vector.push(numbers[n] + numbers[n+1])
    }

    // Sum up 
    let new_sum = new_vector.iter().sum::<i32>();

    sum_of_triangles(&mut new_vector, new_sum)
}

fn main() {

    let mut numbers = vec![1, 2, 3, 4, 5];
    let current_sum = numbers.iter().sum::<i32>();

    let answer3 = sum_of_triangles(&mut numbers, current_sum);
    println!("This is the answer {}", answer3);
}

Sorting

Selection Sort

Selection sort is when we iterate each position and try and get our smallest value each time. E.g. in python

def selection_sort(array):
    m = len(array)
    for i in range(m - 1):
        # Assume the current index (i) has the minimum value
        min_index = i
        # Find the minimum element in the remaining unsorted part
        for j in range(i + 1, m):
            if array[j] < array[min_index]:
                min_index = j
        # Swap the found minimum element with the first element of the unsorted part
        array[i], array[min_index] = array[min_index], array[i]

Looking at the code there are two for loops meaning this is O(n²)
And here it is in rust

fn selection_sort(numbers: &mut Vec<i32>) {
    for i in 0..numbers.len() {
        let mut min_index = i;
        for j in i..numbers.len() {
            if numbers[j] < numbers[min_index] {
                min_index = j;
            }
        }
        numbers.swap(i, min_index);
    }
}

Bubble Sort

This was the first sort I did 40 years ago. Here in rust. Again this is this is O(n²)

fn bubble_sort(numbers: &mut Vec<i32>) {
    for i in 0..numbers.len() {
        for j in 0..numbers.len() - 1 {
            if numbers[j] > numbers[j + 1] {
                numbers.swap(j, j + 1);
            }
        }
    }
}

There were improvements made in the course to the original answer where they identified code repeated even thought we knew the data was already sorted. A clear reason not to just cut and paste.

fn bubble_sort(numbers: &mut Vec<i32>) {
    let mut sorted = true;
    for i in 0..numbers.len() {
        sorted = true;
        for j in 0..numbers.len() - 1 {
            if numbers[j] > numbers[j + 1] {
                numbers.swap(j, j + 1);
                sorted = false;
            }
        }
    }
    
    if sorted {
        break; 
    }
}

Merge Sort

This is more interesting. This is known as a divide and conquer approach as we break the problem up. For this we need to

• Break the array in half until each level is 1 or 2 numbers (4 levels in this picture)
• Create empty array of original length
• Take element either from left of right depending on which is the largest

fn merge_sort(arr: &mut [i32]) -> Vec<i32> {

    // Base case
    if arr.len() > 1 {

        // Split the array into two, left and right
        let mid = arr.len() / 2;
        let left_merge = merge_sort(&mut arr[..mid]);
        let right_merge = merge_sort(&mut arr[mid..]);

        let mut left_index = 0;
        let mut right_index = 0;

        for val in &mut *arr {
            // Determine if left or right index is next
            if right_index == right_merge.len() || (left_index < left_merge.len() && left_merge[left_index] < right_merge[right_index]) {
                // Set left index as next value
                *val = left_merge[left_index];
                left_index += 1;
            } else {
                // Set right index as next value
                *val = right_merge[right_index];
                right_index += 1;
            }
        }
    }

    arr.to_vec()

}

fn main() {
    let mut merge_numbers = vec![38, 27, 43, 3, 9,82, 10];
    let answer4 = merge_sort(&mut merge_numbers);
    println!("This is the answer {:?}", answer4);
}

Quick Sort

A similar approach to sorting but this time we

1. Selection of the Pivot Element: The first step in the QuickSort algorithm is choosing a pivot element from the array. This can be any element from the array.
2. Partitioning: The array is partitioned or reordered so that all elements less than the pivot come before the pivot, while all elements greater than the pivot come after it. After this step, the pivot is in its final position.
3. Recursive Sort: The steps above are recursively applied to the sub-array of elements with smaller values and separately to the sub-array of elements with greater values.

Here is the approach from co pilot

fn quick_sort(array: &mut [i32]) -> Vec<i32> {
    let length = array.len();
    if length <= 1 {
        return array.to_vec();
    }
    let pivot = array[length - 1];
    let mut left = 0;
    let mut right = length - 1;
    while left < right {
        while array[left] < pivot {
            left += 1;
        }
        while array[right] > pivot {
            if right == 0 {
                break;
            }
            right -= 1;
        }
        if left < right {
            array.swap(left, right);
        }
    }
    let mut left_sorted = quick_sort(&mut array[..left]);
    let mut right_sorted = quick_sort(&mut array[left..]);
    left_sorted.append(&mut right_sorted);
    left_sorted

}

The solution by the course was this.

fn quick_sort_2(arr: &mut [i32], start: usize, end: usize) -> Vec<i32> {
    if start < end {
        // Partition function to return the pivot index
        let mut i = start;
        for j in start..end {
            if arr[j] < arr[end] {
                arr.swap(i, j);
                i += 1;
            }
        }

        arr.swap(i, end);
        let p = i;
    
        quick_sort_2(arr, start, p - 1);
        quick_sort_2(arr, p + 1, end);
    }

    arr.to_vec()
}

Macros

I did macros using YouTube and the course. The first example is the course, the second is YouTube.

Greatest Common Denominator Macro

Here is the simple example.

macro_rules! gcd {
     ($a: expr, $b: expr) => {
        {
            let mut a = $a;
            let mut b = $b;
            while b != 0 {
                let t = b;
                b = a % b;
                a = t;
            }
            a
        }
     };
 }

main() {
  print!("gcd is {} ", gcd!(14,28));
}

Declarative Macros

This seems similar to C++ with improvements. Need to learn the syntax but this should be enough for me to get going. Here is a basic macro.

#[macro_export]
macro_rules! my_macro {
    () => {
        { // Do Stuff }
    };
}

Export allows the macro to be seen, macro_rules! is the syntax to say I am a macro then the name of the macro. We can now write an arrow function similar to typescript. This has curly braces around whatever we are returning.
This example appends a "# " to the text passed. The function arguments are $var and in this case the type is literal. These are known as fragment specifiers. And a list can be found [here]

#[macro_export]
macro_rules! header {
    ($var:literal) => {
        {concat!("# ", $var)}
    };
}

fn main() {
    let fred = header!("fred");
    println!("Hello, world! {}", fred);
}

We can test this using the testing framework offered by rust and running cargo test

#[cfg(test)]
mod tests {
    #[test]
    fn test_header() {
        let val = header!("Hello");

        assert_eq!(val, "# Hello"); 
    }
}

A more complicated macro maybe one that takes a list of values. E.g. ulist!["a","b","c"]. First we define a parameter with multiple inputs, to do this start with $(), separator , and repetition operator +. A list of repetition operators may be found [here]

macro_rules! ulist {
   ($(),+) => {
   };
}

To emphasise the separator I have changed it to a % to make it clear how it is used.

#[macro_export]
macro_rules! ulist {
    ($($var:literal)%+) => {
       { concat!( $("item:", $var, "\n",)+ ).trim_end() }
    };
 }

Basically the example concatenates the values together and using .trim_end() removes the last "\n". Left it in as it may remind me how to do this. And here is the test.

    #[test]
    fn test_ulist() {
        let val = ulist!("foo"%"bar");

        assert_eq!(val, "item:foo\nitem:bar");
    }

Procedural Macros

The previous macros were like C++ macros which match against patterns. Procedural macros are like functions, they take code as input, operate on it and then output code. There are three types of these

• Custom #[derive] macros.
• Attribute-like macros.
• Function-like macros

We are going to look at the derived as this is what I have come across and need to know
This type of macro takes code as input and outputs code.

• Create a library crate called billprod_macro
• Create a procedural macro crate inside billprod_macro (billprod_macro_derive)

Here is the code structure.

Macro Library

This just defines the trait we are going to build. In the lib.rs we declare this trait (interface)

pub trait BillProdMacro {
    fn bill_prod_macro();
}

Macro Derive Library

This lives inside of the macro library. For the procedural macro we need to define the Cargo.toml as a proc-macro and add dependencies syn and quote.

[package]
name = "billprod_macro_derive"
version = "0.1.0"
edition = "2021"

# See more keys and their definitions at https://doc.rust-lang.org/cargo/reference/manifest.html
[lib]
proc-macro = true

[dependencies]
syn = "1.0.75"
quote = "1.0.9"

For the code, we have something called TokenStream which is capable of parsing rust code. We write a function to receive the code, do something, and a function to output the code. My understanding is these are the same for most macros. We annotate it with the name of the macro #[proc_macro_derive(BillProdMacro)]

extern crate proc_macro;

use proc_macro::TokenStream;
use quote::quote;
use syn;

#[proc_macro_derive(BillProdMacro)]
pub fn billprod_macro_derive(input: TokenStream) -> TokenStream {

    // Construct a representation of Rust code as a syntax tree
    // that we can manipulate
    let ast = syn::parse(input).unwrap();

    // Build the trait implementation
    impl_billprod_macro(&ast)
}

The implementation takes the code (ast) prints out text with the type.

fn impl_billprod_macro(ast: &syn::DeriveInput) -> TokenStream {
    let name = &ast.ident;
    let gen = quote! {
        impl BillProdMacro for #name {
            fn bill_prod_macro() {
                println!("Hello, world! My name is {}", stringify!(#name));
            }
        }
    };
    gen.into()
}

Using the Macro

We need both packages. As they are not published we must reference them on disk

...
[dependencies]
billprod_macro = { path = "../billprod_macro" }
billprod_macro_derive = { path = "../billprod_macro/billprod_macro_derive" }

So now we can use them in the code.

use billprod_macro::BillProdMacro;
use billprod_macro_derive::BillProdMacro;

#[derive(BillProdMacro)]
struct BillCake;
....

fn main() {
    BillCake::bill_prod_macro();
}

This outputs

Hello, world! My name is BillCake

Implementing Your Types

Another letsgetrusty. This adviced when make a library to play nice by implmenting

• Debug, Clone, Default, PartialEq

It suggested you consider also implementing

• JSON serialize ,deserialize

This code showed me how to optionalize it.

use std::{rc::Rc, sync::Arc};

#[cfg(feature = "serde")]
use serde::{Deserialize, Serialize};

#[derive(Debug, Clone, Default, PartialEq)]
#[cfg_attr(feature = "serde", derive(Serialize, Deserialize))]
pub enum Role {
    Admin,
    Standard,
    #[default]
    Guest,
}

#[derive(Debug, Clone, Default, PartialEq)]
#[cfg_attr(feature = "serde", derive(Serialize, Deserialize))]
struct DB {}

#[derive(Debug, Clone, Default, PartialEq)]
#[cfg_attr(feature = "serde", derive(Serialize, Deserialize))]
pub struct User {
    id: u32,
    name: String,
    role: Role,
    #[cfg_attr(feature = "serde", serde(skip))]
    db: Arc<DB>,
}

fn main() {
    let user = User {
        id: 123,
        name: "Bogdan".to_owned(),
        role: Role::Admin,
        db: Arc::new(DB {}),
    };

    println!("{:?}", user);

    let user2 = user.clone();

    println!("{:?}", user2);

    let guest = User::default();

    let guest2 = User::default();

    assert_eq!(guest, guest2);

    let user_str = "{ \"id\": 123, \"name\": \"Bogdan\", \"role\": \"Admin\" }";

    #[cfg(feature = "serde")]
    let user: User = serde_json::from_str(&user_str).unwrap();

    #[cfg(feature = "serde")]
    println!("{:?}", user);
}

fn is_normal<T: Sized + Send + Sync + Unpin>() {}

#[test]
fn normal_types() {
    is_normal::<User>();
}

And here is the toml

[package]
name = "common-traits"
version = "0.1.0"
edition = "2021"

# See more keys and their definitions at https://doc.rust-lang.org/cargo/reference/manifest.html

[dependencies]
serde = { version = "1.0", features = ["derive", "rc"], optional = true }
serde_json = "1.0"

[features]
serde = ["dep:serde"]

WASM

Web assembly for rust is done via

Functions

Functions and Arguments

No surprises

fn print_value(x: i32)
{
  println("value = {}", x);
}

Pass by reference

fn increase(x: &mut i32)
{
  *x = 1;
}

Return value, note no semicolon

fn product(x: i32, y: i32) -> i32
{
  x * y
}

Return two values, note no semicolon

fn product(x: i32) -> i32
{
  if x == 10 {
     123
  }
  
  321
}

Higher-order functions

Not sure what this is, seems like just a way to chain written functions together like lamba. Here is the given example.

fn is_even(x: u32) -> bool 
{
   x%2 == 0 
}

fn main()
{
// Method without HOF
  let limit = 500;
  let mut sum = 0;

  for i in 0..
  {
    let isq = i*i;

    if isq > limit { break; }
    else if is_even(isq) { sum += isq; }
  }

  println!("loop sum = {}", sum);

// HOF way

  let sum2 =
     (0...).map(|x| x*x)
        .take_while(|&x| x < limit)
        .filter(|x| is_even(*x))
        .fold(0, |sum,x| sum+x);

  println!("hof sum = {}", sum2);
}

Odds and Ends

Consuming Crates

Crates is like nuget [1]

[package]
name = "mypackage" 
version = "0.1.0"
author = " ["Iain Wiseman iwiseman@bibble.co.nz"]

[dependencies]
rand = "0.3.12"

And the usage

extern crate rand; // Package
use rand::Rng; // Namespace

fn main() {
  let mut rng = rand::thread_rng();
  let b:bool = rng.gen();
}

Building Crates and Modules

Module example e.g. src/lib.rs

pub mod greetings
{
   pub mod english
   {
      pub fn hello() -> {"hello"->to_string() }
      pub fn goodbye() -> {"goodbye"->to_string() }
   }
   pub mod french
   {
      pub fn hello() -> {"bonjour"->to_string() }
      pub fn goodbye() -> {"au revoir"->to_string() }
   }
}

For modules within modules you can make directories. For above this would be greetings\english and greetings\french. So we could have greetings\english\english.rs

  pub fn hello() -> {"hello"->to_string() }
  pub fn goodbye() -> {"goodbye"->to_string() }

greetings\lib.rs

pub mod greetings
{
   pub mod english;
   pub mod french
   {
      pub fn hello() -> {"bonjour"->to_string() }
      pub fn goodbye() -> {"au revoir"->to_string() }
   }
}

We need a package file for it

[package]
name = "phrases" 
version = "0.1.0"
author = " ["Iain Wiseman iwiseman@bibble.co.nz"]

And then cargo build should work. To use the package you will need to specify the path to the package in the cargo file where used. e.g.

[package]
name = "mypackage" 
version = "0.1.0"
author = " ["Iain Wiseman iwiseman@bibble.co.nz"]

[dependencies]
rand = "0.3.12"
phrases = { path = "../Phrases" }

Testing

Rush comes with it's own assert library marked by #[test] e.g. for the above

#[test]
fn english_greeting_correct()
{
   assert_eq!("hello", greetings::english::hello());
}

Documentation

rustdoc is used to generate.

//! This module comment
//! and this
//! #Examples are compiled with back ticks 
//! ```
//! let username = "John";
//! println!("{}", english::hello());
//! ```

/// This is for code
/// In this case, it's our `hello()` function.
pub fn hello() -> String {" hello".to_string() }

Installing

Do not use apt as it does not set rustup correctly and then vscode extension will not work with "Couldn't start client Rust Language Server"

apt-get install curl build-essential make gcc -y
curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh

VS Code setup

I ended up installing. Still cannot find a in built package manager for cargo

Date Handling

Found the on twitter (not all bad) @orhanbalci Type conversion and dates are the worst in most languages so thought I would keep this.

fn main() {
    // you can use chrono crate for date time operations.
    // chrono crate is more capable than standard library std::time module
    use chrono::prelude::*;
    use chrono::Duration;
    
    // you can retrieve current date in Utc timezone as follows
    let current_date = Utc::today();
    println!("Utc current date: {}", current_date);
    
    // you can retrieve current date in local timezone as follows
    let local_current_date = Local::today();
    println!("Local current date: {}", local_current_date);
    
    // you can retrieve current time in UTC as follows
    let current_time_utc = Utc::now();
    println!( "Utc current time: {}", current_time_utc);
    
    // you can retrieve current time in local time zone as follows
    let current_time_local = Local::now();
    println!("Local current time: {}", current_time_local);
    
    // you can add some duration to a chrono::Date
    // succ method gets succeeding date
    
    let today = Utc::today();
    
    let tomorrow = today + Duration::days(1);
    assert_eq!(today.succ(), tomorrow);
    
    // you can subtract some duration from a chrono
    // pred method gets previous date
    
    let today = Utc::today();
    
    let yesterday = today - Duration::days(1);
    assert_eq!(today.pred(), yesterday);
    
    // you can get UNIX timestamp (epoch) value for a datetime using
    
    // timestamp method of chrono::offset::TimeZone trait.
    
    // since timestamps have numeric representation, they are easy to store in db
    // and send through network. You can aldo prefer this notation in your APIs
    let dt = Utc.ymd(1970, 1, 1).and_hms_milli(0, 0, 1, 0);
    assert_eq!(dt.timestamp(), 1);
    
    //you can also get timestamp value of a datetime in milliseconds
    let dt = Utc.ymd(1970, 1, 1).and_hms_milli(0, 0, 1, 500);
    assert_eq!(dt.timestamp_millis(), 1500);
    
    //you can convert create a chrono::DateTime from a timestamp seconds
    let timestamp = 15;
    
    let datetime = Utc.timestamp(timestamp, 0);
    assert_eq!(datetime.timestamp(), 15);
    
    //you can get difference of two date times as follows
    let first = Utc.ymd(1970, 1, 1).and_hms_milli(0, 0, 1, 0);
    let second = Utc.ymd(1970, 1, 1).and_hms_milli(0, 0, 2, 0);
    
    let difference: Duration = second.signed_duration_since(first);
    assert_eq!(difference, Duration::seconds(1));
    
    // you can also add and subtract duration from a DateTime struct
    
    let now = Utc::now();
    
    let yesterday_at_the_same_time = now - Duration::days(1);
    println!("Yesterday at the same time: {}", yesterday_at_the_same_time);
    
    //you can compare durations
    assert_eq! (Duration::days(1), Duration::hours(24));
    
    //you can format your DateTime struct using strftime formatting options
    let now = Utc::now();
    println!("{}", now. format( "%d.%m.%Y %H:%M"));
    
    //you can parse DateTime from a formatted string as below
    let some_time = NaiveDateTime::parse_from_str("01.10.2021 14:21", "%d.%m.%Y %H:%M").unwrap( );
    assert_eq!(some_time, NaiveDate::from_ymd(2021,10,1).and_hms(14,21,0));
}

Type State Pattern

In some cases we only want to be able to do some functions on a struct based on the state of the struct. In rust we could create a field in the struct called state and maintain which functions are allowed based on the current state. However rust provides a better approach referred to as the type state pattern. For example if we wanted make a password manager structure we could do this.

#![allow(unused)]

use std::collections::HashMap;
use std::marker::PhantomData;

struct Locked;
struct Unlocked;

// PasswordManager<Locked> != PasswordManager<Unlocked>

struct PasswordManager<State = Locked> {
    master_pass: String,
    passwords: HashMap<String, String>,
    state: PhantomData<State>,
}

// Locked Functions
impl PasswordManager<Locked> {
    pub fn unlock(self, master_pass: String) -> PasswordManager<Unlocked> {
        PasswordManager {
            master_pass: self.master_pass,
            passwords: self.passwords,
            state: PhantomData,
        }
    }
}

// Unlocked Functions
impl PasswordManager<Unlocked> {
    pub fn lock(self) -> PasswordManager<Locked> {
        PasswordManager {
            master_pass: self.master_pass,
            passwords: self.passwords,
            state: PhantomData,
        }
    }

    pub fn list_passwords(&self) -> &HashMap<String, String> {
        &self.passwords
    }

    pub fn add_password(&mut self, username: String, password: String) {
        self.passwords.insert(username, password);
    }
}

// Generic Functions
impl<State> PasswordManager<State> {
    pub fn encryption(&self) -> String {
        todo!()
    }

    pub fn version(&self) -> String {
        todo!()
    }
}

// Constructor
impl PasswordManager {
    pub fn new(master_pass: String) -> Self {
        PasswordManager {
            master_pass,
            passwords: Default::default(),
            state: PhantomData,
        }
    }
}

fn main() {
    let mut manager = PasswordManager::new("password123".to_owned());
    let manager = manager.unlock("password123".to_owned());
    manager.list_passwords();
    manager.lock();
}

The benefits to doing this are

• Only functions allowed are displayed in the IDE. E.g. a locked password manager cannot see the functions for an unlocked password manager
• The state PhantomData<State> is compiled out and reduces memory footprint