543 lines
20 KiB
Rust
Raw Normal View History

2021-08-04 07:55:54 +08:00
use crate::{
core::{attributes::*, operation::*},
errors::OTError,
};
2021-07-31 20:53:45 +08:00
use bytecount::num_chars;
2021-08-03 23:13:20 +08:00
use std::{cmp::Ordering, fmt, iter::FromIterator, str::FromStr};
2021-07-31 20:53:45 +08:00
#[derive(Clone, Debug, PartialEq)]
pub struct Delta {
pub ops: Vec<Operation>,
pub base_len: usize,
pub target_len: usize,
}
impl Default for Delta {
fn default() -> Self {
Self {
ops: Vec::new(),
base_len: 0,
target_len: 0,
}
}
}
2021-08-01 10:45:15 +08:00
impl FromStr for Delta {
type Err = ();
fn from_str(s: &str) -> Result<Delta, Self::Err> {
let mut delta = Delta::with_capacity(1);
delta.add(Operation::Insert(s.into()));
Ok(delta)
}
}
impl<T: AsRef<str>> From<T> for Delta {
fn from(s: T) -> Delta { Delta::from_str(s.as_ref()).unwrap() }
}
2021-08-03 23:13:20 +08:00
impl fmt::Display for Delta {
fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
f.write_str(&serde_json::to_string(self).unwrap_or("".to_owned()))?;
// for op in &self.ops {
// f.write_fmt(format_args!("{}", op));
// }
Ok(())
}
}
2021-08-01 10:45:15 +08:00
impl FromIterator<Operation> for Delta {
fn from_iter<T: IntoIterator<Item = Operation>>(ops: T) -> Self {
let mut operations = Delta::default();
for op in ops {
operations.add(op);
}
operations
}
}
2021-07-31 20:53:45 +08:00
impl Delta {
#[inline]
pub fn with_capacity(capacity: usize) -> Self {
Self {
ops: Vec::with_capacity(capacity),
base_len: 0,
target_len: 0,
}
}
2021-08-01 14:39:30 +08:00
pub fn add(&mut self, op: Operation) {
2021-08-01 10:45:15 +08:00
match op {
Operation::Delete(i) => self.delete(i),
2021-08-01 14:39:30 +08:00
Operation::Insert(i) => self.insert(&i.s, i.attributes),
Operation::Retain(r) => self.retain(r.num, r.attributes),
2021-08-01 10:45:15 +08:00
}
}
2021-07-31 20:53:45 +08:00
pub fn delete(&mut self, n: u64) {
if n == 0 {
return;
}
self.base_len += n as usize;
2021-08-01 10:45:15 +08:00
if let Some(Operation::Delete(n_last)) = self.ops.last_mut() {
*n_last += n;
} else {
self.ops.push(OpBuilder::delete(n).build());
2021-07-31 20:53:45 +08:00
}
}
2021-08-03 16:20:25 +08:00
pub fn insert(&mut self, s: &str, attrs: Attributes) {
2021-07-31 20:53:45 +08:00
if s.is_empty() {
return;
}
2021-08-01 10:45:15 +08:00
2021-08-01 16:39:32 +08:00
self.target_len += num_chars(s.as_bytes());
2021-08-01 10:45:15 +08:00
let new_last = match self.ops.as_mut_slice() {
[.., Operation::Insert(insert)] => {
2021-08-01 16:39:32 +08:00
//
merge_insert_or_new_op(insert, s, attrs)
2021-07-31 20:53:45 +08:00
},
[.., Operation::Insert(pre_insert), Operation::Delete(_)] => {
2021-08-01 16:39:32 +08:00
//
merge_insert_or_new_op(pre_insert, s, attrs)
2021-07-31 20:53:45 +08:00
},
2021-08-01 10:45:15 +08:00
[.., op_last @ Operation::Delete(_)] => {
2021-07-31 20:53:45 +08:00
let new_last = op_last.clone();
2021-08-01 16:39:32 +08:00
*op_last = OpBuilder::insert(s).attributes(attrs).build();
Some(new_last)
2021-07-31 20:53:45 +08:00
},
2021-08-01 16:39:32 +08:00
_ => Some(OpBuilder::insert(s).attributes(attrs).build()),
2021-07-31 20:53:45 +08:00
};
2021-08-01 16:39:32 +08:00
match new_last {
None => {},
Some(new_last) => self.ops.push(new_last),
}
2021-07-31 20:53:45 +08:00
}
2021-08-03 16:20:25 +08:00
pub fn retain(&mut self, n: u64, attrs: Attributes) {
2021-07-31 20:53:45 +08:00
if n == 0 {
return;
}
self.base_len += n as usize;
self.target_len += n as usize;
if let Some(Operation::Retain(retain)) = self.ops.last_mut() {
match merge_retain_or_new_op(retain, n, attrs) {
2021-08-01 16:39:32 +08:00
None => {},
Some(new_op) => self.ops.push(new_op),
}
2021-08-01 10:45:15 +08:00
} else {
self.ops
2021-08-01 14:39:30 +08:00
.push(OpBuilder::retain(n).attributes(attrs).build());
2021-07-31 20:53:45 +08:00
}
}
/// Merges the operation with `other` into one operation while preserving
/// the changes of both. Or, in other words, for each input string S and a
/// pair of consecutive operations A and B.
/// `apply(apply(S, A), B) = apply(S, compose(A, B))`
/// must hold.
///
/// # Error
///
/// Returns an `OTError` if the operations are not composable due to length
/// conflicts.
pub fn compose(&self, other: &Self) -> Result<Self, OTError> {
if self.target_len != other.base_len {
return Err(OTError);
}
let mut new_delta = Delta::default();
let mut ops1 = self.ops.iter().cloned();
let mut ops2 = other.ops.iter().cloned();
2021-08-01 10:45:15 +08:00
let mut next_op1 = ops1.next();
let mut next_op2 = ops2.next();
2021-07-31 20:53:45 +08:00
loop {
2021-08-01 10:45:15 +08:00
match (&next_op1, &next_op2) {
2021-07-31 20:53:45 +08:00
(None, None) => break,
2021-08-01 10:45:15 +08:00
(Some(Operation::Delete(i)), _) => {
2021-07-31 20:53:45 +08:00
new_delta.delete(*i);
2021-08-01 10:45:15 +08:00
next_op1 = ops1.next();
2021-07-31 20:53:45 +08:00
},
(_, Some(Operation::Insert(o_insert))) => {
2021-08-03 16:20:25 +08:00
new_delta.insert(
&o_insert.s,
attributes_from(&next_op2).unwrap_or(Attributes::Empty),
);
2021-08-01 10:45:15 +08:00
next_op2 = ops2.next();
2021-07-31 20:53:45 +08:00
},
(None, _) | (_, None) => {
return Err(OTError);
},
(Some(Operation::Retain(retain)), Some(Operation::Retain(o_retain))) => {
2021-08-03 16:20:25 +08:00
let composed_attrs = compose_attributes(&next_op1, &next_op2);
2021-08-02 23:08:24 +08:00
log::debug!(
"[retain:{} - retain:{}]: {:?}",
retain.num,
o_retain.num,
composed_attrs
);
match retain.cmp(&o_retain) {
2021-07-31 23:22:17 +08:00
Ordering::Less => {
new_delta.retain(retain.num, composed_attrs);
next_op2 = Some(OpBuilder::retain(o_retain.num - retain.num).build());
2021-08-01 10:45:15 +08:00
next_op1 = ops1.next();
2021-07-31 23:22:17 +08:00
},
std::cmp::Ordering::Equal => {
new_delta.retain(retain.num, composed_attrs);
2021-08-01 10:45:15 +08:00
next_op1 = ops1.next();
next_op2 = ops2.next();
2021-07-31 23:22:17 +08:00
},
std::cmp::Ordering::Greater => {
new_delta.retain(o_retain.num, composed_attrs);
next_op1 = Some(OpBuilder::retain(retain.num - o_retain.num).build());
2021-08-01 10:45:15 +08:00
next_op2 = ops2.next();
2021-07-31 23:22:17 +08:00
},
}
2021-07-31 20:53:45 +08:00
},
(Some(Operation::Insert(insert)), Some(Operation::Delete(o_num))) => {
match (num_chars(insert.as_bytes()) as u64).cmp(o_num) {
2021-07-31 20:53:45 +08:00
Ordering::Less => {
2021-08-01 10:45:15 +08:00
next_op2 = Some(
OpBuilder::delete(*o_num - num_chars(insert.as_bytes()) as u64)
2021-08-03 16:20:25 +08:00
.attributes(insert.attributes.clone())
.build(),
2021-07-31 20:53:45 +08:00
);
2021-08-01 10:45:15 +08:00
next_op1 = ops1.next();
2021-07-31 20:53:45 +08:00
},
Ordering::Equal => {
2021-08-01 10:45:15 +08:00
next_op1 = ops1.next();
next_op2 = ops2.next();
2021-07-31 20:53:45 +08:00
},
Ordering::Greater => {
2021-08-01 10:45:15 +08:00
next_op1 = Some(
2021-08-01 16:39:32 +08:00
OpBuilder::insert(
&insert.chars().skip(*o_num as usize).collect::<String>(),
2021-08-01 16:39:32 +08:00
)
.build(),
2021-07-31 20:53:45 +08:00
);
2021-08-01 10:45:15 +08:00
next_op2 = ops2.next();
2021-07-31 20:53:45 +08:00
},
}
},
(Some(Operation::Insert(insert)), Some(Operation::Retain(o_retain))) => {
2021-08-03 16:20:25 +08:00
let composed_attrs = compose_attributes(&next_op1, &next_op2);
2021-08-02 23:08:24 +08:00
log::debug!(
"[insert:{} - retain:{}]: {:?}",
insert.s,
o_retain.num,
composed_attrs
);
match (insert.num_chars()).cmp(o_retain) {
2021-07-31 20:53:45 +08:00
Ordering::Less => {
new_delta.insert(&insert.s, composed_attrs.clone());
next_op2 = Some(
OpBuilder::retain(o_retain.num - insert.num_chars())
.attributes(composed_attrs.clone())
.build(),
);
2021-08-01 10:45:15 +08:00
next_op1 = ops1.next();
2021-07-31 20:53:45 +08:00
},
Ordering::Equal => {
new_delta.insert(&insert.s, composed_attrs);
2021-08-01 10:45:15 +08:00
next_op1 = ops1.next();
next_op2 = ops2.next();
2021-07-31 20:53:45 +08:00
},
Ordering::Greater => {
2021-08-01 10:45:15 +08:00
let chars = &mut insert.chars();
new_delta.insert(
&chars.take(o_retain.num as usize).collect::<String>(),
composed_attrs,
);
2021-08-03 16:20:25 +08:00
next_op1 = Some(
OpBuilder::insert(&chars.collect::<String>())
.attributes(Attributes::Empty)
.build(),
);
2021-08-01 10:45:15 +08:00
next_op2 = ops2.next();
2021-07-31 20:53:45 +08:00
},
}
},
(Some(Operation::Retain(retain)), Some(Operation::Delete(o_num))) => {
match retain.cmp(&o_num) {
Ordering::Less => {
new_delta.delete(retain.num);
next_op2 = Some(OpBuilder::delete(*o_num - retain.num).build());
next_op1 = ops1.next();
},
Ordering::Equal => {
new_delta.delete(*o_num);
next_op2 = ops2.next();
next_op1 = ops1.next();
},
Ordering::Greater => {
new_delta.delete(*o_num);
next_op1 = Some(OpBuilder::retain(retain.num - *o_num).build());
next_op2 = ops2.next();
},
}
2021-07-31 20:53:45 +08:00
},
};
}
Ok(new_delta)
}
/// Transforms two operations A and B that happened concurrently and
/// produces two operations A' and B' (in an array) such that
/// `apply(apply(S, A), B') = apply(apply(S, B), A')`.
/// This function is the heart of OT.
///
/// # Error
///
/// Returns an `OTError` if the operations cannot be transformed due to
/// length conflicts.
pub fn transform(&self, other: &Self) -> Result<(Self, Self), OTError> {
if self.base_len != other.base_len {
return Err(OTError);
}
let mut a_prime = Delta::default();
let mut b_prime = Delta::default();
let mut ops1 = self.ops.iter().cloned();
let mut ops2 = other.ops.iter().cloned();
2021-08-01 10:45:15 +08:00
let mut next_op1 = ops1.next();
let mut next_op2 = ops2.next();
2021-07-31 20:53:45 +08:00
loop {
2021-08-01 10:45:15 +08:00
match (&next_op1, &next_op2) {
2021-07-31 20:53:45 +08:00
(None, None) => break,
2021-08-01 10:45:15 +08:00
(Some(Operation::Insert(insert)), _) => {
// let composed_attrs = transform_attributes(&next_op1, &next_op2, true);
a_prime.insert(&insert.s, insert.attributes.clone());
b_prime.retain(insert.num_chars(), insert.attributes.clone());
2021-08-01 10:45:15 +08:00
next_op1 = ops1.next();
2021-07-31 20:53:45 +08:00
},
(_, Some(Operation::Insert(o_insert))) => {
let composed_attrs = transform_attributes(&next_op1, &next_op2, true);
a_prime.retain(o_insert.num_chars(), composed_attrs.clone());
2021-08-03 16:20:25 +08:00
b_prime.insert(&o_insert.s, composed_attrs);
2021-08-01 10:45:15 +08:00
next_op2 = ops2.next();
2021-07-31 20:53:45 +08:00
},
(None, _) => {
return Err(OTError);
},
(_, None) => {
return Err(OTError);
},
(Some(Operation::Retain(retain)), Some(Operation::Retain(o_retain))) => {
let composed_attrs = transform_attributes(&next_op1, &next_op2, true);
match retain.cmp(&o_retain) {
2021-07-31 20:53:45 +08:00
Ordering::Less => {
a_prime.retain(retain.num, composed_attrs.clone());
b_prime.retain(retain.num, composed_attrs.clone());
next_op2 = Some(OpBuilder::retain(o_retain.num - retain.num).build());
2021-08-01 10:45:15 +08:00
next_op1 = ops1.next();
2021-07-31 20:53:45 +08:00
},
Ordering::Equal => {
a_prime.retain(retain.num, composed_attrs.clone());
b_prime.retain(retain.num, composed_attrs.clone());
2021-08-01 10:45:15 +08:00
next_op1 = ops1.next();
next_op2 = ops2.next();
2021-07-31 20:53:45 +08:00
},
Ordering::Greater => {
a_prime.retain(o_retain.num, composed_attrs.clone());
b_prime.retain(o_retain.num, composed_attrs.clone());
next_op1 = Some(OpBuilder::retain(retain.num - o_retain.num).build());
2021-08-01 10:45:15 +08:00
next_op2 = ops2.next();
2021-07-31 20:53:45 +08:00
},
};
},
2021-08-01 10:45:15 +08:00
(Some(Operation::Delete(i)), Some(Operation::Delete(j))) => match i.cmp(&j) {
2021-07-31 20:53:45 +08:00
Ordering::Less => {
2021-08-01 10:45:15 +08:00
next_op2 = Some(OpBuilder::delete(*j - *i).build());
next_op1 = ops1.next();
2021-07-31 20:53:45 +08:00
},
Ordering::Equal => {
2021-08-01 10:45:15 +08:00
next_op1 = ops1.next();
next_op2 = ops2.next();
2021-07-31 20:53:45 +08:00
},
Ordering::Greater => {
2021-08-01 10:45:15 +08:00
next_op1 = Some(OpBuilder::delete(*i - *j).build());
next_op2 = ops2.next();
2021-07-31 20:53:45 +08:00
},
},
(Some(Operation::Delete(i)), Some(Operation::Retain(o_retain))) => {
match i.cmp(&o_retain) {
2021-07-31 20:53:45 +08:00
Ordering::Less => {
a_prime.delete(*i);
next_op2 = Some(OpBuilder::retain(o_retain.num - *i).build());
2021-08-01 10:45:15 +08:00
next_op1 = ops1.next();
2021-07-31 20:53:45 +08:00
},
Ordering::Equal => {
a_prime.delete(*i);
2021-08-01 10:45:15 +08:00
next_op1 = ops1.next();
next_op2 = ops2.next();
2021-07-31 20:53:45 +08:00
},
Ordering::Greater => {
a_prime.delete(o_retain.num);
next_op1 = Some(OpBuilder::delete(*i - o_retain.num).build());
2021-08-01 10:45:15 +08:00
next_op2 = ops2.next();
2021-07-31 20:53:45 +08:00
},
};
},
(Some(Operation::Retain(retain)), Some(Operation::Delete(j))) => {
match retain.cmp(&j) {
2021-07-31 20:53:45 +08:00
Ordering::Less => {
b_prime.delete(retain.num);
next_op2 = Some(OpBuilder::delete(*j - retain.num).build());
2021-08-01 10:45:15 +08:00
next_op1 = ops1.next();
2021-07-31 20:53:45 +08:00
},
Ordering::Equal => {
b_prime.delete(retain.num);
2021-08-01 10:45:15 +08:00
next_op1 = ops1.next();
next_op2 = ops2.next();
2021-07-31 20:53:45 +08:00
},
Ordering::Greater => {
b_prime.delete(*j);
next_op1 = Some(OpBuilder::retain(retain.num - *j).build());
2021-08-01 10:45:15 +08:00
next_op2 = ops2.next();
2021-07-31 20:53:45 +08:00
},
};
},
}
}
Ok((a_prime, b_prime))
}
/// Applies an operation to a string, returning a new string.
///
/// # Error
///
/// Returns an error if the operation cannot be applied due to length
/// conflicts.
pub fn apply(&self, s: &str) -> Result<String, OTError> {
if num_chars(s.as_bytes()) != self.base_len {
return Err(OTError);
}
let mut new_s = String::new();
let chars = &mut s.chars();
for op in &self.ops {
2021-08-01 10:45:15 +08:00
match &op {
Operation::Retain(retain) => {
for c in chars.take(retain.num as usize) {
2021-07-31 20:53:45 +08:00
new_s.push(c);
}
},
2021-08-01 10:45:15 +08:00
Operation::Delete(delete) => {
2021-07-31 20:53:45 +08:00
for _ in 0..*delete {
chars.next();
}
},
2021-08-01 10:45:15 +08:00
Operation::Insert(insert) => {
new_s += &insert.s;
2021-07-31 20:53:45 +08:00
},
}
}
Ok(new_s)
}
/// Computes the inverse of an operation. The inverse of an operation is the
/// operation that reverts the effects of the operation
pub fn invert(&self, s: &str) -> Self {
2021-08-01 10:45:15 +08:00
let mut inverted = Delta::default();
2021-07-31 20:53:45 +08:00
let chars = &mut s.chars();
for op in &self.ops {
2021-08-01 10:45:15 +08:00
match &op {
Operation::Retain(retain) => {
2021-08-03 16:20:25 +08:00
inverted.retain(retain.num, Attributes::Follow);
for _ in 0..retain.num {
2021-07-31 20:53:45 +08:00
chars.next();
}
},
2021-08-01 10:45:15 +08:00
Operation::Insert(insert) => {
inverted.delete(insert.num_chars());
2021-07-31 20:53:45 +08:00
},
2021-08-01 10:45:15 +08:00
Operation::Delete(delete) => {
inverted.insert(
2021-07-31 23:22:17 +08:00
&chars.take(*delete as usize).collect::<String>(),
op.get_attributes(),
2021-07-31 23:22:17 +08:00
);
2021-07-31 20:53:45 +08:00
},
}
}
2021-08-01 10:45:15 +08:00
inverted
2021-07-31 20:53:45 +08:00
}
/// Checks if this operation has no effect.
#[inline]
pub fn is_noop(&self) -> bool {
2021-08-01 10:45:15 +08:00
match self.ops.as_slice() {
2021-07-31 20:53:45 +08:00
[] => true,
2021-08-01 10:45:15 +08:00
[Operation::Retain(_)] => true,
2021-07-31 20:53:45 +08:00
_ => false,
}
}
/// Returns the length of a string these operations can be applied to
#[inline]
pub fn base_len(&self) -> usize { self.base_len }
/// Returns the length of the resulting string after the operations have
/// been applied.
#[inline]
pub fn target_len(&self) -> usize { self.target_len }
/// Returns the wrapped sequence of operations.
#[inline]
pub fn ops(&self) -> &[Operation] { &self.ops }
2021-08-01 10:45:15 +08:00
pub fn is_empty(&self) -> bool { self.ops.is_empty() }
2021-07-31 20:53:45 +08:00
}
2021-07-31 23:22:17 +08:00
2021-08-01 16:39:32 +08:00
fn merge_insert_or_new_op(
insert: &mut Insert,
s: &str,
2021-08-03 16:20:25 +08:00
attributes: Attributes,
2021-08-01 16:39:32 +08:00
) -> Option<Operation> {
2021-08-03 16:20:25 +08:00
match &attributes {
Attributes::Follow => {
insert.s += s;
return None;
},
Attributes::Custom(_) | Attributes::Empty => {
if insert.attributes == attributes {
insert.s += s;
None
} else {
Some(OpBuilder::insert(s).attributes(attributes).build())
}
},
2021-08-01 16:39:32 +08:00
}
}
fn merge_retain_or_new_op(
retain: &mut Retain,
n: u64,
2021-08-03 16:20:25 +08:00
attributes: Attributes,
2021-08-01 16:39:32 +08:00
) -> Option<Operation> {
2021-08-03 16:20:25 +08:00
log::debug!(
2021-08-02 23:08:24 +08:00
"merge_retain_or_new_op: {:?}, {:?}",
retain.attributes,
attributes
);
2021-08-03 16:20:25 +08:00
match &attributes {
Attributes::Follow => {
retain.num += n;
None
},
Attributes::Custom(_) | Attributes::Empty => {
if retain.attributes == attributes {
retain.num += n;
None
} else {
Some(OpBuilder::retain(n).attributes(attributes).build())
}
},
2021-07-31 23:22:17 +08:00
}
}