Compare commits

..

2 Commits

4 changed files with 285 additions and 5 deletions

View File

@ -15,5 +15,5 @@ tui = { version = "0.19", default-features = false, features = [ "crosst
[[bin]]
name = "bench"
main = "bin/bench.rs"
features = [ "bench" ]
# main = "bin/bench.rs"
# features = [ "bench" ]

View File

@ -0,0 +1,269 @@
use std::{mem, ops::Deref};
const CHUNK_SIZE: usize = 10;
#[derive(Clone, Copy, Debug)]
struct NodeId(usize);
impl Deref for NodeId {
type Target = usize;
fn deref(&self) -> &Self::Target {
&self.0
}
}
#[derive(Debug)]
enum Node {
Interior {
// the number of characters to the left of this node
// the number of lines to the left of this node
char_count: usize,
left: Option<NodeId>,
right: Option<NodeId>,
},
Leaf(LeafNode),
}
impl Node {
fn len(&self) -> usize {
match self {
Node::Interior { char_count, .. } => 0,
Node::Leaf(ln) => ln.len(),
}
}
}
#[derive(Debug)]
struct LeafNode(String);
impl LeafNode {
fn len(&self) -> usize {
self.0.len()
}
fn take(&mut self) -> String {
let content = mem::replace(&mut self.0, "".to_owned());
content
}
fn as_str(&self) -> &str {
&self.0
}
}
impl From<String> for LeafNode {
fn from(s: String) -> Self {
Self(s)
}
}
/// A Rope is a regular tree which adds on some extra behavior for dealing with a continuous data
/// structure. In this case, the nodes all contain strings, and the rope is arranged such that a
/// depth-first traversal will yield the entire contents of the rope in proper order.
#[derive(Debug)]
pub struct Rope {
node_count: usize,
contents: Vec<Option<Node>>,
}
impl Rope {
/// Insert text at an index within the document. loc is the number of characters from the
/// beginning.
pub fn insert_at(&mut self, loc: usize, text: String) {
match self.find_insertion_node_id(loc) {
None => {
let node = Node::Leaf(LeafNode::from(text));
self.node_count += 1;
self.contents.push(Some(node));
}
Some(id) => {
self.insert_at_node(id, text);
}
}
}
/// Append text to the end of the document.
pub fn append(&mut self, text: String) {
self.insert_at(self.len(), text);
}
/// Convert the entire rope back to a continuous String.
pub fn to_string(&self) -> String {
if self.contents.is_empty() {
return "".to_owned();
}
let mut r = String::new();
let mut stack = vec![NodeId(0)];
while let Some(current_id) = stack.pop() {
let node = &self.contents[*current_id];
match node {
Some(Node::Interior { left, right, .. }) => {
if let Some(right_id) = *right {
stack.push(right_id);
}
if let Some(left_id) = *left {
stack.push(left_id);
}
}
Some(Node::Leaf(ln)) => r.push_str(ln.as_str()),
None => panic!("Should never leave an empty space in the node list"),
}
}
r
}
/// Calculate the length of the stored string.
pub fn len(&self) -> usize {
// This can be optimized later. Do a traversal of each right node. We already have
// character counts of each left tree. Only count the length of the final right leaf.
self.contents.iter().fold(0, |acc, node| {
if let Some(Node::Leaf(s)) = node {
acc + s.len()
} else {
acc
}
})
}
#[cfg(test)]
fn max_depth(&self) -> usize {
unimplemented!();
}
#[cfg(test)]
fn node_count(&self) -> usize {
self.node_count
}
// Find the node ID of the insertion point. This is not fully implemented, in that this
// function ignores the offset from the beginning. Because of that, it is also always inserting
// onto the right side, and never traversing down the left.
fn find_insertion_node_id(&self, _loc: usize) -> Option<NodeId> {
let mut current_id = NodeId(0);
loop {
match self.contents.get(*current_id) {
Some(Some(Node::Interior { ref right, .. })) => match right {
Some(id) => current_id = *id,
None => return Some(current_id),
},
Some(Some(Node::Leaf(_))) => return Some(current_id),
Some(None) => panic!("There should never be an empty node in the tree"),
// This only happens when the list is empty. Otherwise, we're detecting the None in
// advance.
None => return None,
}
}
}
// Insert text at a particular node location.
//
// This is not a self-balancing operation (yet). Once we know where text needs to be inserted,
// based on the offset from the beginning, we can grab that node and either replace it (if it
// is a Leaf node) or update it (if it is an Interior node).
//
// This function is currently naive, in that it will always assume that text needs to be added
// to the right side, which may not be correct.
fn insert_at_node(&mut self, id: NodeId, text: String) {
match self.contents[*id] {
Some(Node::Interior { ref mut right, .. }) => {
let new_node = Node::Leaf(LeafNode::from(text));
let new_node_id = NodeId(self.node_count + 1);
*right = Some(new_node_id);
self.contents.push(Some(new_node));
self.node_count += 1;
}
Some(Node::Leaf(_)) => {
let Some(Node::Leaf(mut ln)) = mem::replace(&mut self.contents[*id], None) else {
panic!("Should never leave an empty space in the node list")
};
let contents = ln.take();
let lnode = Node::Leaf(LeafNode::from(contents));
let rnode = Node::Leaf(LeafNode::from(text));
let lnode_id = self.node_count;
let rnode_id = self.node_count + 1;
let interior_node = Node::Interior {
char_count: lnode.len(),
left: Some(NodeId(lnode_id)),
right: Some(NodeId(rnode_id)),
};
let _ = mem::replace(&mut self.contents[*id], Some(interior_node));
self.node_count += 2;
self.contents.push(Some(lnode));
self.contents.push(Some(rnode));
}
None => panic!("Should never leave an empty space in the node list"),
}
}
}
impl Default for Rope {
fn default() -> Self {
Self {
node_count: 0,
contents: vec![],
}
}
}
// Populate the initial rope. The simplest way is to split along lines and turn each line into its
// own leaf node.
impl From<String> for Rope {
fn from(s: String) -> Self {
let mut rope = Rope::default();
#[allow(unused_assignments)]
let mut first = s.as_str();
let mut lst = s.as_str();
while lst.len() > CHUNK_SIZE {
(first, lst) = lst.split_at(CHUNK_SIZE);
rope.append(first.to_owned());
}
rope.append(lst.to_owned());
rope
}
}
#[cfg(test)]
mod test {
use super::*;
struct TestCase {
content: String,
}
#[test]
fn it_creates_a_rope_from_a_string() {
let test_cases = vec![
TestCase{ content: "".to_owned() },
TestCase{ content: "This".to_owned() },
TestCase{ content: "This is some basic".to_owned() },
TestCase{ content: "This is some basic context which is much smaller".to_owned() },
TestCase{ content:
"This is some basic context which is much smaller than the rope is designed for."
.to_owned()
},
];
for case in test_cases {
let rope = Rope::from(case.content.clone());
for (idx, node) in rope.contents.iter().enumerate() {
}
assert_eq!(rope.len(), case.content.len(), "{}", case.content);
assert_eq!(rope.to_string(), case.content, "{:?}", case.content);
}
}
}

View File

@ -1,3 +1,4 @@
pub mod doc_rope;
pub mod ui;
pub mod state;
pub mod types;

View File

@ -14,6 +14,10 @@ impl Document {
Self { rows: contents }
}
pub fn line(&self, id: usize) -> Option<&str> {
self.rows.get(id).map(|x| x.as_str())
}
pub fn contents(&self) -> String {
self.rows.join("\n")
}
@ -117,7 +121,7 @@ mod test_utils {
time::{Duration, Instant},
};
pub fn with_file<F>(test: F)
pub fn with_moby_dick<F>(test: F)
where
F: FnOnce(Document),
{
@ -163,12 +167,18 @@ mod test {
#[test]
fn it_inserts_a_line() {
with_file(|mut doc| {
with_moby_dick(|mut doc| {
let mut cursor = Cursor::default();
let num_lines = doc.row_count();
assert_eq!(
doc.line(num_lines - 3),
Some("subscribe to our email newsletter to hear about new eBooks.")
);
doc.new_line(&mut cursor);
assert_eq!(doc.row_count(), num_lines + 1);
assert_eq!(doc.line(0), Some(""));
});
}
}
@ -177,7 +187,7 @@ pub mod bench {
use super::{test_utils::*, *};
pub fn bench_insert_lines() {
with_file(|doc| {
with_moby_dick(|doc| {
let performance = benchmark(
1000,
|| doc.clone(),