Merge pull request #3 from upupnoah/main

feat(resp-decode): support resp decode for simple redis

Cargo.toml

@@ -10,3 +10,4 @@ authors = ["Noah <[email protected]>"]
 anyhow = "^1.0"
 bytes = "^1.6.1"
 enum_dispatch = "^0.3.13"
+thiserror = "^1.0.62"

examples/bytes_mut.rs

@@ -0,0 +1,15 @@
+use anyhow::Result;
+use bytes::BytesMut;
+fn main() -> Result<()> {
+    let a = "hello"; // a.as_bytes() = b"hello"
+
+    // bytes_mut
+    let mut bytes_mut = BytesMut::new();
+    bytes_mut.extend_from_slice(a.as_bytes());
+    println!("bytes_mut: {:?}", bytes_mut);
+
+    let b = bytes_mut.split_to(3);
+    println!("b: {:?}", b);
+    println!("after split_to(3) -> bytes_mut: {:?}", bytes_mut);
+    Ok(())
+}

examples/enum_dispatch.rs

@@ -0,0 +1,39 @@
+use enum_dispatch::enum_dispatch;
+
+#[enum_dispatch]
+trait DoSomething {
+    fn do_something(&self);
+}
+
+#[enum_dispatch(DoSomething)]
+enum Types {
+    Apple(A),
+    Banana(B),
+}
+
+struct A;
+struct B;
+
+impl DoSomething for A {
+    fn do_something(&self) {
+        println!("A");
+    }
+}
+
+impl DoSomething for B {
+    fn do_something(&self) {
+        println!("B");
+    }
+}
+fn main() {
+    // test enum_dispatch
+    let apple = Types::Apple(A);
+    let banana = Types::Banana(B);
+
+    let type_apple = apple;
+    let type_banana = banana;
+
+    // 都是 types 类型的, 但是结果不同, enum_dispatch 相当于是主动帮我 match 了
+    type_apple.do_something();
+    type_banana.do_something();
+}

examples/thiserorr_anyhow.rs

@@ -0,0 +1,57 @@
+use anyhow::{Context, Result};
+use std::io::Error as IoError;
+use std::num::ParseIntError;
+use thiserror::Error;
+
+// 定义自定义错误类型
+#[derive(Error, Debug)]
+enum MyError {
+    #[error("An IO error occurred: {0}")]
+    Io(#[from] IoError),
+
+    #[error("A parsing error occurred: {0}")]
+    Parse(#[from] ParseIntError),
+
+    #[error("Custom error: {0}")]
+    Custom(String),
+
+    #[error("Anyhow error: {0}")]
+    Anyhow(#[from] anyhow::Error),
+}
+
+// 一个可能返回错误的函数
+fn parse_number(input: &str) -> Result<i32, MyError> {
+    let trimmed = input.trim();
+    if trimmed.is_empty() {
+        return Err(MyError::Custom("Input is empty".into()));
+    }
+
+    let number: i32 = trimmed
+        .parse()
+        // .map_err(|e| MyError::Parse(e))
+        .map_err(MyError::Parse) // 更好的写法
+        .context("Failed to parse number")?;
+    Ok(number)
+}
+
+fn main() -> Result<(), MyError> {
+    // 示例一: 正确的输入
+    match parse_number("42") {
+        Ok(number) => println!("Parsed number: {}", number),
+        Err(e) => eprintln!("Error: {}", e),
+    }
+
+    // 示例二: 空输入
+    match parse_number("") {
+        Ok(number) => println!("Parsed number: {}", number),
+        Err(e) => eprintln!("Error: {}", e),
+    }
+
+    // 示例三: 无效输入
+    match parse_number("abc") {
+        Ok(number) => println!("Parsed number: {}", number),
+        Err(e) => eprintln!("Error: {}", e),
+    }
+
+    Ok(())
+}

src/resp.rs

@@ -1,18 +1,60 @@
-use std::collections::BTreeMap;
-
-use bytes::BytesMut;
+use bytes::{Buf, BytesMut};
 use enum_dispatch::enum_dispatch;
+use std::collections::BTreeMap;
+use thiserror::Error;
 
+mod decode;
 mod encode;
 
+const CRLF: &[u8] = b"\r\n";
+const CRLF_LEN: usize = CRLF.len();
+
 #[enum_dispatch]
 pub trait RespEncode {
     fn encode(self) -> Vec<u8>;
 }
 
-pub trait RespDecode {
-    fn decode(buf: Self) -> Result<RespFrame, String>;
+// Sized 表示这个 trait 只能被 [大小确定的类型] 实现
+// 因为 decode 方法的返回值是一个 Self, 因此必须将这个 trait 标记为 Sized
+pub trait RespDecode: Sized {
+    const PREFIX: &'static str;
+    fn decode(buf: &mut BytesMut) -> Result<Self, RespError>;
+    fn expect_length(buf: &[u8]) -> Result<usize, RespError>;
+}
+
+#[derive(Error, Debug, PartialEq, Eq)]
+pub enum RespError {
+    // region:    --- thiserror format usage
+
+    // #[error("{var}")] ⟶ write!("{}", self.var)
+    // #[error("{0}")] ⟶ write!("{}", self.0)
+    // #[error("{var:?}")] ⟶ write!("{:?}", self.var)
+    // #[error("{0:?}")] ⟶ write!("{:?}", self.0)
+
+    // endregion: --- thiserror format usage
+    #[error("Invalid frame: {0}")] // 这里的 0 表示 self.0。 会转化为 write!
+    InvalidFrame(String),
+    #[error("Invalid frame type: {0}")]
+    InvalidFrameType(String),
+    #[error("Invalid frame length: {0}")]
+    InvalidFrameLength(isize),
+    #[error("Frame is not complete")]
+    NotComplete,
+
+    #[error("Parse error: {0}")]
+    ParseIntError(#[from] std::num::ParseIntError),
+    #[error("Utf8 error: {0}")]
+    Utf8Error(#[from] std::string::FromUtf8Error),
+    #[error("Parse float error: {0}")]
+    ParseFloatError(#[from] std::num::ParseFloatError),
 }
+
+// pub trait RespDecode: Sized {
+//     const PREFIX: &'static str;
+//     fn decode(buf: &mut BytesMut) -> Result<Self, RespError>;
+//     fn expect_length(buf: &[u8]) -> Result<usize, RespError>;
+// }
+
 // 之所以要定义一些新的结构体, 是因为要在实现 trait 的时候, 要区分开这些类型
 #[enum_dispatch(RespEncode)]
 pub enum RespFrame {
@@ -35,12 +77,19 @@ pub enum RespFrame {
 // 2. 新类型模式：这是 Rust 中常用的一种模式，用于在类型系统层面区分不同用途的相同底层类型。比如，你可能想区分普通的字符串和特定格式的字符串。
 // 3. 添加方法：你可以为 SimpleString 实现方法，这些方法特定于这种类型的字符串。
 // 4. 语义清晰：在复杂的数据结构中（如你展示的 RespFrame 枚举），使用 SimpleString 而不是直接使用 String 可以使代码的意图更加明确。
+#[derive(Debug, Clone, PartialEq, Eq, PartialOrd)]
 pub struct SimpleString(String); // Simple String, 用于存储简单字符串
+#[derive(Debug, Clone, PartialEq, Eq, PartialOrd)]
 pub struct SimpleError(String);
+#[derive(Debug, Clone, PartialEq, Eq, PartialOrd)]
 pub struct BulkString(Vec<u8>); // 单个二进制字符串, 用于存储二进制数据(最大512MB)
+#[derive(Debug, Clone, PartialEq, Eq, PartialOrd)]
 pub struct RespNullBulkString;
+
 pub struct RespArray(Vec<RespFrame>);
+#[derive(Debug, Clone, PartialEq, Eq, PartialOrd)]
 pub struct RespNullArray;
+#[derive(Debug, Clone, PartialEq, Eq, PartialOrd)]
 pub struct RespNull;
 #[derive(Default)]
 pub struct RespMap(BTreeMap<String, RespFrame>); // 改为 BTreeMap, 用于有序的 key-value 数据
@@ -84,14 +133,91 @@ impl RespSet {
     }
 }
 
-impl RespDecode for BytesMut {
-    fn decode(_buf: Self) -> Result<RespFrame, String> {
-        todo!()
+// utility functions
+fn extract_fixed_data(
+    buf: &mut BytesMut,
+    expect: &str,
+    expect_type: &str,
+) -> Result<(), RespError> {
+    if buf.len() < expect.len() {
+        return Err(RespError::NotComplete);
     }
+
+    if !buf.starts_with(expect.as_bytes()) {
+        return Err(RespError::InvalidFrameType(format!(
+            "expect: {}, got: {:?}",
+            expect_type, buf
+        )));
+    }
+
+    buf.advance(expect.len());
+    Ok(())
 }
 
-// impl RespEncode for RespFrame {
-//     fn encode(self) -> Vec<u8> {
-//         todo!()
-//     }
-// }
+fn extract_simple_frame_data(buf: &[u8], prefix: &str) -> Result<usize, RespError> {
+    if buf.len() < 3 {
+        return Err(RespError::NotComplete);
+    }
+
+    if !buf.starts_with(prefix.as_bytes()) {
+        return Err(RespError::InvalidFrameType(format!(
+            "expect: SimpleString({}), got: {:?}",
+            prefix, buf
+        )));
+    }
+
+    let end = find_crlf(buf, 1).ok_or(RespError::NotComplete)?;
+
+    Ok(end)
+}
+
+// find nth CRLF in the buffer
+fn find_crlf(buf: &[u8], nth: usize) -> Option<usize> {
+    let mut count = 0;
+    for i in 1..buf.len() - 1 {
+        if buf[i] == b'\r' && buf[i + 1] == b'\n' {
+            count += 1;
+            if count == nth {
+                return Some(i);
+            }
+        }
+    }
+    None
+}
+
+fn parse_length(buf: &[u8], prefix: &str) -> Result<(usize, usize), RespError> {
+    let end = extract_simple_frame_data(buf, prefix)?;
+    let s = String::from_utf8_lossy(&buf[prefix.len()..end]);
+    Ok((end, s.parse()?))
+}
+
+fn calc_total_length(buf: &[u8], end: usize, len: usize, prefix: &str) -> Result<usize, RespError> {
+    let mut total = end + CRLF_LEN;
+    let mut data = &buf[total..];
+    match prefix {
+        "*" | "~" => {
+            // find nth CRLF in the buffer, for array and set, we need to find 1 CRLF for each element
+            for _ in 0..len {
+                let len = RespFrame::expect_length(data)?;
+                data = &data[len..];
+                total += len;
+            }
+            Ok(total)
+        }
+        "%" => {
+            // find nth CRLF in the buffer. For map, we need to find 2 CRLF for each key-value pair
+            for _ in 0..len {
+                let len = SimpleString::expect_length(data)?;
+
+                data = &data[len..];
+                total += len;
+
+                let len = RespFrame::expect_length(data)?;
+                data = &data[len..];
+                total += len;
+            }
+            Ok(total)
+        }
+        _ => Ok(len + CRLF_LEN),
+    }
+}