Syncing with codecrafters-io/languages

Created by https://github.com/codecrafters-io/languages

.gitignore

@@ -0,0 +1,12 @@
+# Generated by Cargo
+# will have compiled files and executables
+debug/
+target/
+
+# These are backup files generated by rustfmt
+**/*.rs.bk
+
+# MSVC Windows builds of rustc generate these, which store debugging information
+*.pdb
+
+*.db

Cargo.lock

@@ -0,0 +1,14 @@
+# This file is automatically @generated by Cargo.
+# It is not intended for manual editing.
+[[package]]
+name = "anyhow"
+version = "1.0.44"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "61604a8f862e1d5c3229fdd78f8b02c68dcf73a4c4b05fd636d12240aaa242c1"
+
+[[package]]
+name = "sqlite-starter-rust"
+version = "0.1.0"
+dependencies = [
+ "anyhow",
+]

Cargo.toml

@@ -0,0 +1,22 @@
+# DON'T EDIT THIS!
+#
+# Codecrafters relies on this file being intact to run tests successfully. Any changes
+# here will not reflect when CodeCrafters tests your code, and might even cause build
+# failures.
+#
+# DON'T EDIT THIS!
+[package]
+name = "sqlite-starter-rust"
+version = "0.1.0"
+authors = ["Codecrafters <hello@codecrafters.io>"]
+edition = "2018"
+
+# DON'T EDIT THIS!
+#
+# Codecrafters relies on this file being intact to run tests successfully. Any changes
+# here will not reflect when CodeCrafters tests your code, and might even cause build
+# failures.
+#
+# DON'T EDIT THIS!
+[dependencies]
+anyhow = "1.0.43" # for easy error handling

README.md

@@ -1 +1,85 @@
-# sqlite-starter-rust
+This is a starting point for Rust solutions to the
+["Build Your Own SQLite" Challenge](https://codecrafters.io/challenges/sqlite).
+
+In this challenge, you'll build a barebones SQLite implementation that supports
+basic SQL queries like `SELECT`. Along the way we'll learn about
+[SQLite's file format](https://www.sqlite.org/fileformat.html), how indexed data
+is
+[stored in B-trees](https://jvns.ca/blog/2014/10/02/how-does-sqlite-work-part-2-btrees/)
+and more.
+
+**Note**: If you're viewing this repo on GitHub, head over to
+[codecrafters.io](https://codecrafters.io) to signup for early access.
+
+# Passing the first stage
+
+CodeCrafters runs tests when you do a `git push`. Make an empty commit and push
+your solution to see the first stage fail.
+
+```sh
+git commit --allow-empty -m "Running tests"
+git push origin master
+```
+
+You should see a failure message that says it expected "number of tables: <n>"
+to be present.
+
+Go to `src/main.rs` and uncomment the `.dbinfo` command implementation. Commit
+and push your changes to pass the first stage:
+
+```sh
+git add .
+git commit -m "pass the first stage"
+git push origin master
+```
+
+Time to move on to the next stage!
+
+# Running Your Program Locally
+
+1. Ensure you have `cargo (1.43)` installed locally
+1. Run `./your_sqlite3.sh` to run your program, which is implemented in
+   `src/main.rs`. This command compiles your Rust project, so it might be slow
+   the first time you run it. Subsequent runs will be fast.
+1. Commit your changes and run `git push origin master` to submit your solution
+   to CodeCrafters. Test output will be streamed to your terminal.
+
+# Sample Databases
+
+To make it easy to test queries locally, we've added a sample database in the
+root of this repository: `sample.db`.
+
+This contains two tables: `apples` & `oranges`. You can use this to test your
+implementation for the first 6 stages.
+
+You can explore this database by running queries against it like this:
+
+```sh
+$ sqlite3 sample.db "select id, name from apples"
+1|Granny Smith
+2|Fuji
+3|Honeycrisp
+4|Golden Delicious
+```
+
+There are two other databases that you can use:
+
+1. `superheroes.db`:
+   - This is a small version of the test database used in the table-scan stage.
+   - It contains one table: `superheroes`.
+   - It is ~1MB in size.
+1. `companies.db`:
+   - This is a small version of the test database used in the index-scan stage.
+   - It contains one table: `companies`, and one index: `idx_companies_country`
+   - It is ~7MB in size.
+
+These aren't included in the repository because they're large in size. You can
+download them by running this script:
+
+```sh
+./download_sample_databases.sh
+```
+
+If the script doesn't work for some reason, you can download the databases
+directly from
+[codecrafters-io/sample-sqlite-databases](https://github.com/codecrafters-io/sample-sqlite-databases).

codecrafters.yml

@@ -0,0 +1,11 @@
+# Set this to true if you want debug logs.
+#
+# These can be VERY verbose, so we suggest turning them off
+# unless you really need them.
+debug: false
+
+# Use this to change the Rust version used to run your code
+# on Codecrafters.
+#
+# Available versions: rust-1.43
+language_pack: rust-1.43

download_sample_databases.sh

@@ -0,0 +1,9 @@
+#!/bin/sh
+
+echo "Downloading superheroes.db: ~1MB (used in stage 7)"
+curl -Lo superheroes.db https://raw.githubusercontent.com/codecrafters-io/sample-sqlite-databases/master/superheroes.db
+
+echo "Downloading companies.db: ~7MB (used in stage 8)"
+curl -Lo companies.db https://raw.githubusercontent.com/codecrafters-io/sample-sqlite-databases/master/companies.db
+
+echo "Sample databases downloaded."

src/header.rs

@@ -0,0 +1,44 @@
+use anyhow::{bail, Result};
+use std::convert::TryInto;
+
+#[derive(Debug)]
+pub enum BTreePage {
+    InteriorIndex = 2,
+    InteriorTable = 5,
+    LeafIndex = 10,
+    LeafTable = 13,
+}
+
+#[derive(Debug)]
+pub struct PageHeader {
+    pub page_type: BTreePage,
+    pub first_free_block_start: u16,
+    pub number_of_cells: u16,
+    pub start_of_content_area: u16,
+    pub fragmented_free_bytes: u8,
+}
+
+impl PageHeader {
+    /// Parses a page header stream into a page header
+    pub fn parse(stream: &[u8]) -> Result<Self> {
+        let page_type = match stream[0] {
+            2 => BTreePage::InteriorIndex,
+            5 => BTreePage::InteriorTable,
+            10 => BTreePage::LeafIndex,
+            13 => BTreePage::LeafTable,
+            x => bail!("Invalid page value encountered: {}", x),
+        };
+        let first_free_block_start = u16::from_be_bytes(stream[1..3].try_into()?);
+        let number_of_cells = u16::from_be_bytes(stream[3..5].try_into()?);
+        let start_of_content_area = u16::from_be_bytes(stream[5..7].try_into()?);
+        let fragmented_free_bytes = stream[7];
+        let header = PageHeader {
+            page_type,
+            first_free_block_start,
+            number_of_cells,
+            start_of_content_area,
+            fragmented_free_bytes,
+        };
+        Ok(header)
+    }
+}

src/lib.rs

@@ -0,0 +1,4 @@
+pub mod header;
+pub mod record;
+pub mod schema;
+pub mod varint;

src/main.rs

@@ -0,0 +1,59 @@
+use anyhow::{bail, Result};
+use sqlite_starter_rust::{
+    header::PageHeader, record::parse_record, schema::Schema, varint::parse_varint,
+};
+use std::convert::TryInto;
+use std::fs::File;
+use std::io::prelude::*;
+
+fn main() -> Result<()> {
+    // Parse arguments
+    let args = std::env::args().collect::<Vec<_>>();
+    match args.len() {
+        0 | 1 => bail!("Missing <database path> and <command>"),
+        2 => bail!("Missing <command>"),
+        _ => {}
+    }
+
+    // Read database file into database
+    let mut file = File::open(&args[1])?;
+    let mut database = Vec::new();
+    file.read_to_end(&mut database)?;
+
+    // Parse command and act accordingly
+    let command = &args[2];
+    match command.as_str() {
+        ".dbinfo" => {
+            // Parse page header from database
+            let page_header = PageHeader::parse(&database[100..108])?;
+
+            // Obtain all cell pointers
+            let cell_pointers = database[108..]
+                .chunks_exact(2)
+                .take(page_header.number_of_cells.into())
+                .map(|bytes| u16::from_be_bytes(bytes.try_into().unwrap()))
+                .collect::<Vec<_>>();
+
+            // Obtain all records from column 5
+            #[allow(unused_variables)]
+            let schemas = cell_pointers
+                .into_iter()
+                .map(|cell_pointer| {
+                    let stream = &database[cell_pointer as usize..];
+                    let (_, offset) = parse_varint(stream);
+                    let (_rowid, read_bytes) = parse_varint(&stream[offset..]);
+                    parse_record(&stream[offset + read_bytes..], 5)
+                        .map(|record| Schema::parse(record).expect("Invalid record"))
+                })
+                .collect::<Result<Vec<_>>>()?;
+
+            println!("Your code goes here!");
+
+            // Uncomment this block to pass the first stage
+            // println!("number of tables: {}", schemas.len());
+        }
+        _ => bail!("Missing or invalid command passed: {}", command),
+    }
+
+    Ok(())
+}

src/record.rs

@@ -0,0 +1,42 @@
+use crate::varint::parse_varint;
+use anyhow::{bail, Result};
+
+/// Reads SQLite's "Record Format" as mentioned here:
+/// [record_format](https://www.sqlite.org/fileformat.html#record_format)
+pub fn parse_record(stream: &[u8], column_count: usize) -> Result<Vec<Vec<u8>>> {
+    // Parse number of bytes in header, and use bytes_read as offset
+    let (_, mut offset) = parse_varint(stream);
+
+    // Read each varint into serial types and modify the offset
+    let mut serial_types = vec![];
+    for _ in 0..column_count {
+        let (varint, read_bytes) = parse_varint(&stream[offset..]);
+        offset += read_bytes;
+        serial_types.push(varint);
+    }
+
+    // Parse each serial type as column into record and modify the offset
+    let mut record = vec![];
+    for serial_type in serial_types {
+        let column = parse_column_value(&stream[offset..], serial_type as u8)?;
+        offset += column.len();
+        record.push(column);
+    }
+
+    Ok(record)
+}
+
+fn parse_column_value(stream: &[u8], serial_type: u8) -> Result<Vec<u8>> {
+    let column_value = match serial_type {
+        // 8 bit twos-complement integer
+        1 => vec![stream[0]],
+        // Text encoding
+        n if serial_type >= 13 && serial_type % 2 == 1 => {
+            let n_bytes = (n - 13) / 2;
+            let bytes = stream[0..n_bytes as usize].to_vec();
+            bytes
+        }
+        _ => bail!("Invalid serial_type: {}", serial_type),
+    };
+    Ok(column_value)
+}

src/schema.rs

@@ -0,0 +1,29 @@
+#[derive(Debug)]
+pub struct Schema {
+    kind: String,
+    name: String,
+    table_name: String,
+    root_page: u8,
+    sql: String,
+}
+
+impl Schema {
+    /// Parses a record into a schema
+    pub fn parse(record: Vec<Vec<u8>>) -> Option<Self> {
+        let mut items = record.into_iter();
+        let kind = items.next()?;
+        let name = items.next()?;
+        let table_name = items.next()?;
+        let root_page = *items.next()?.get(0)?;
+        let sql = items.next()?;
+
+        let schema = Self {
+            kind: String::from_utf8_lossy(&kind).to_string(),
+            name: String::from_utf8_lossy(&name).to_string(),
+            table_name: String::from_utf8_lossy(&table_name).to_string(),
+            root_page,
+            sql: String::from_utf8_lossy(&sql).to_string(),
+        };
+        Some(schema)
+    }
+}

src/varint.rs

@@ -0,0 +1,46 @@
+const IS_FIRST_BIT_ZERO_MASK: u8 = 0b10000000;
+const LAST_SEVEN_BITS_MASK: u8 = 0b01111111;
+
+/// Parses SQLite's "varint" (short for variable-length integer) as mentioned here:
+/// [varint](https://www.sqlite.org/fileformat2.html#varint)
+///
+/// Returns (varint, bytes_read)
+pub fn parse_varint(stream: &[u8]) -> (usize, usize) {
+    let usable_bytes = read_usable_bytes(stream);
+    let bytes_read = usable_bytes.len();
+    let varint = usable_bytes
+        .into_iter()
+        .enumerate()
+        .fold(0, |value, (i, usable_byte)| {
+            let usable_size = if i == 8 { 8 } else { 7 };
+            (value << usable_size) + usable_value(usable_size, usable_byte) as usize
+        });
+    (varint, bytes_read)
+}
+
+/// Usable size is either 8 or 7
+fn usable_value(usable_size: u8, byte: u8) -> u8 {
+    if usable_size == 8 {
+        usable_size
+    } else {
+        byte & LAST_SEVEN_BITS_MASK
+    }
+}
+
+fn read_usable_bytes(stream: &[u8]) -> Vec<u8> {
+    let mut usable_bytes = vec![];
+
+    for i in 0..8 {
+        let byte = stream[i];
+        usable_bytes.push(byte);
+        if starts_with_zero(byte) {
+            break;
+        }
+    }
+
+    usable_bytes
+}
+
+fn starts_with_zero(byte: u8) -> bool {
+    (byte & IS_FIRST_BIT_ZERO_MASK) == 0
+}

your_sqlite3.sh

@@ -0,0 +1,12 @@
+#!/bin/sh
+#
+# DON'T EDIT THIS!
+#
+# CodeCrafters uses this file to test your code. Don't make any changes here!
+#
+# DON'T EDIT THIS!
+exec cargo run \
+    --quiet \
+    --release \
+    --target-dir=/tmp/codecrafters-sqlite-target \
+    --manifest-path "$(dirname "$0")/Cargo.toml" "$@"