Reworked diff algorithm to work with streams

This commit is contained in:
FyloZ 2023-06-28 23:12:19 -04:00
parent 93d460ae5d
commit d44616bd72
Signed by: william
GPG Key ID: 835378AE9AF4AE97
4 changed files with 253 additions and 88 deletions

View File

@ -1,7 +1,8 @@
// Based on https://github.com/mathertel/Diff // Based on https://github.com/mathertel/Diff
// "An O(ND) Difference Algorithm and its Variations" by Eugene Myers Algorithmica Vol. 1 No. 2, 1986, p 251. // "An O(ND) Difference Algorithm and its Variations" by Eugene Myers Algorithmica Vol. 1 No. 2, 1986, p 251.
use std::collections::HashMap; use std::collections::HashMap;
use std::io::{BufRead};
use crate::diff::IndexDirection::{None, LeftDown, RightUp};
#[derive(Debug)] #[derive(Debug)]
pub struct DiffItem { pub struct DiffItem {
@ -22,44 +23,84 @@ struct SmsData {
y: usize, y: usize,
} }
struct DiffVec { struct SmsBounds {
data: Vec<usize>, lower_a: usize,
lower_b: usize,
upper_a: usize,
upper_b: usize,
max_d: usize,
down_k: i32,
up_k: i32,
down_offset: usize,
up_offset: usize,
is_delta_odd: bool,
} }
impl DiffVec { impl SmsBounds {
fn get(&self, index: usize) -> usize { fn from(lower_a: usize, lower_b: usize, upper_a: usize, upper_b: usize) -> Self {
self.data[index] let max = upper_a + upper_b + 1;
} let max_d = (upper_a - lower_a + upper_b - lower_b) / 2 + 1;
fn get_i32(&self, index: i32) -> usize { let down_k = lower_a as i32 - lower_b as i32;
if index < 0 { let up_k = upper_a as i32 - upper_b as i32;
panic!("Got index < 0");
let down_offset = (max as i32 - down_k) as usize;
let up_offset = (max as i32 - up_k) as usize;
let is_delta_odd = ((upper_a + lower_a + upper_b + lower_b) & 1) != 0;
SmsBounds {
lower_a,
lower_b,
upper_a,
upper_b,
max_d,
down_k,
up_k,
down_offset,
up_offset,
is_delta_odd,
} }
self.get(index as usize)
} }
fn set(&mut self, index: usize, val: usize) { fn get_down_index(&self, k: i32, dir: IndexDirection) -> usize {
self.data[index] = val; self.get_index(self.down_offset, k, dir)
} }
fn set_i32(&mut self, index: i32, val: usize) { fn get_up_index(&self, k: i32, dir: IndexDirection) -> usize {
if index < 0 { self.get_index(self.up_offset, k, dir)
panic!("Got index < 0"); }
}
self.set(index as usize, val); fn get_index(&self, offset: usize, k: i32, dir: IndexDirection) -> usize {
(offset as i32 + k + dir as i32) as usize
} }
} }
pub fn diff(a: &str, b: &str) -> Vec<DiffItem> { #[repr(i32)]
let mut existing_hashes: HashMap<&str, usize> = HashMap::new(); enum IndexDirection {
None = 0,
LeftDown = 1, // Down: Down, Up: Left
RightUp = -1, // Down: Up, Up: Right
}
// https://stackoverflow.com/questions/54035728/how-to-add-a-negative-i32-number-to-an-usize-variable
fn add_i32(index: usize, offset: i32) -> usize {
if offset.is_negative() {
index - offset.wrapping_abs() as u32 as usize
} else {
index + offset as usize
}
}
pub fn diff<T>(a: &mut T, b: &mut T) -> Vec<DiffItem>
where T: BufRead {
let mut existing_hashes: HashMap<String, usize> = HashMap::new();
let mut data_a = diff_data(a, &mut existing_hashes); let mut data_a = diff_data(a, &mut existing_hashes);
let mut data_b = diff_data(b, &mut existing_hashes); let mut data_b = diff_data(b, &mut existing_hashes);
let max = data_a.length + data_b.length; let max = data_a.length + data_b.length;
let mut down_vector = DiffVec { data: vec![0usize; 2 * max + 2] }; let mut down_vector = vec![0usize; 2 * max + 2];
let mut up_vector = DiffVec { data: vec![0usize; 2 * max + 2] }; let mut up_vector = vec![0usize; 2 * max + 2];
let upper_a = data_a.length; let upper_a = data_a.length;
let upper_b = data_b.length; let upper_b = data_b.length;
@ -72,8 +113,9 @@ pub fn diff(a: &str, b: &str) -> Vec<DiffItem> {
create_diffs(&data_a, &data_b) create_diffs(&data_a, &data_b)
} }
fn diff_data<'a>(text: &'a str, existing_hashes: &mut HashMap<&'a str, usize>) -> DiffData { fn diff_data<T>(reader: &mut T, existing_hashes: &mut HashMap<String, usize>) -> DiffData
let codes = diff_codes(text, existing_hashes); where T: BufRead {
let codes = diff_codes(reader, existing_hashes);
let length = codes.len(); let length = codes.len();
DiffData { DiffData {
@ -83,21 +125,24 @@ fn diff_data<'a>(text: &'a str, existing_hashes: &mut HashMap<&'a str, usize>) -
} }
} }
fn diff_codes<'a>(text: &'a str, existing_hashes: &mut HashMap<&'a str, usize>) -> Vec<usize> { fn diff_codes<T>(reader: &mut T, existing_hashes: &mut HashMap<String, usize>) -> Vec<usize>
let lines: Vec<&str> = text.split('\n').collect(); where T: BufRead {
let mut codes = Vec::new();
let mut codes = vec![0usize; lines.len()];
let mut next_code = existing_hashes.len() + 1; let mut next_code = existing_hashes.len() + 1;
for i in 0..lines.len() { loop {
let line = lines[i]; let mut line = String::new();
let read_res = reader.read_line(&mut line).expect("Failed to read BufRead");
if read_res == 0 {
break;
}
if !existing_hashes.contains_key(line) { if !existing_hashes.contains_key(&line) {
existing_hashes.insert(line, next_code); existing_hashes.insert(line, next_code);
codes[i] = next_code; codes.push(next_code);
next_code += 1; next_code += 1;
} else { } else {
codes[i] = existing_hashes[line]; codes.push(existing_hashes[&line]);
} }
} }
@ -105,7 +150,7 @@ fn diff_codes<'a>(text: &'a str, existing_hashes: &mut HashMap<&'a str, usize>)
} }
// Longest Common-Subsequence // Longest Common-Subsequence
fn lcs(data_a: &mut DiffData, mut lower_a: usize, mut upper_a: usize, data_b: &mut DiffData, mut lower_b: usize, mut upper_b: usize, down_vector: &mut DiffVec, up_vector: &mut DiffVec) { fn lcs(data_a: &mut DiffData, mut lower_a: usize, mut upper_a: usize, data_b: &mut DiffData, mut lower_b: usize, mut upper_b: usize, down_vector: &mut Vec<usize>, up_vector: &mut Vec<usize>) {
while lower_a < upper_a && lower_b < upper_b && data_a.codes[lower_a] == data_b.codes[lower_b] { while lower_a < upper_a && lower_b < upper_b && data_a.codes[lower_a] == data_b.codes[lower_b] {
lower_a += 1; lower_a += 1;
lower_b += 1; lower_b += 1;
@ -130,7 +175,8 @@ fn lcs(data_a: &mut DiffData, mut lower_a: usize, mut upper_a: usize, data_b: &m
} }
} else { } else {
// Find the middle snake and length of an optimal path for A and B // Find the middle snake and length of an optimal path for A and B
let sms = sms(&data_a, lower_a, upper_a, &data_b, lower_b, upper_b, down_vector, up_vector); let sms_bounds = SmsBounds::from(lower_a, lower_b, upper_a, upper_b);
let sms = sms(&data_a, &data_b, &sms_bounds, down_vector, up_vector);
// The path is from lower_x to (x, y) and (x, y) to upper_x // The path is from lower_x to (x, y) and (x, y) to upper_x
lcs(data_a, lower_a, sms.x, data_b, lower_b, sms.y, down_vector, up_vector); lcs(data_a, lower_a, sms.x, data_b, lower_b, sms.y, down_vector, up_vector);
@ -138,96 +184,77 @@ fn lcs(data_a: &mut DiffData, mut lower_a: usize, mut upper_a: usize, data_b: &m
} }
} }
// https://stackoverflow.com/questions/54035728/how-to-add-a-negative-i32-number-to-an-usize-variable
fn add_i32(index: usize, offset: i32) -> usize {
if offset.is_negative() {
index - offset.wrapping_abs() as u32 as usize
} else {
index + offset as usize
}
}
// Shortest Middle Snake // Shortest Middle Snake
fn sms(data_a: &DiffData, lower_a: usize, upper_a: usize, data_b: &DiffData, lower_b: usize, upper_b: usize, down_vector: &mut DiffVec, up_vector: &mut DiffVec) -> SmsData { fn sms(data_a: &DiffData, data_b: &DiffData, bounds: &SmsBounds, down_vector: &mut Vec<usize>, up_vector: &mut Vec<usize>) -> SmsData {
let max = upper_a as i32 + upper_b as i32 + 1; down_vector[bounds.get_down_index(bounds.down_k, LeftDown)] = bounds.lower_a;
up_vector[bounds.get_up_index(bounds.up_k, RightUp)] = bounds.upper_a;
let down_k = lower_a as i32 - lower_b as i32; for d in 0..=bounds.max_d as i32 {
let up_k = upper_a as i32 - upper_b as i32;
let delta = (upper_a - lower_a) - (upper_b - lower_b);
let odd_delta = (delta & 1) != 0;
let down_offset = max - down_k;
let up_offset = max - up_k;
let max_d = ((upper_a - lower_a + upper_b - lower_b) / 2) as i32 + 1;
down_vector.set_i32(down_offset + down_k + 1, lower_a);
up_vector.set_i32(up_offset + up_k - 1, upper_a);
for d in 0..=max_d {
// Extend the forward path // Extend the forward path
for k in ((down_k - d)..=(down_k + d)).step_by(2) { for k in ((bounds.down_k - d)..=(bounds.down_k + d)).step_by(2) {
let mut x; let mut x;
let mut y; let mut y;
if k == down_k - d {
if k == bounds.down_k - d {
// Down // Down
x = down_vector.get_i32(down_offset + k + 1); x = down_vector[bounds.get_down_index(k, LeftDown)];
} else { } else {
// Right // Right
x = down_vector.get_i32(down_offset + k - 1) + 1; x = down_vector[bounds.get_down_index(k, RightUp)] + 1;
if k < down_k + d && down_vector.get_i32(down_offset + k + 1) >= x { if k < bounds.down_k + d && down_vector[bounds.get_down_index(k, LeftDown)] >= x {
// Down // Down
x = down_vector.get_i32(down_offset + k + 1); x = down_vector[bounds.get_down_index(k, LeftDown)];
} }
} }
y = add_i32(x, -k); y = add_i32(x, -k);
// Find the end of the furthest reaching forward D-path in diagonal k. // Find the end of the furthest reaching forward D-path in diagonal k.
while x < upper_a && y < upper_b && data_a.codes[x] == data_b.codes[y] { while x < bounds.upper_a && y < bounds.upper_b && data_a.codes[x] == data_b.codes[y] {
x += 1; x += 1;
y += 1; y += 1;
} }
down_vector.set_i32(down_offset + k, x); down_vector[bounds.get_down_index(k, None)] = x;
// Overlap ? // Overlap ?
if odd_delta && up_k - d < k && k < up_k + d && up_vector.get_i32(up_offset + k) <= down_vector.get_i32(down_offset + k) { if bounds.is_delta_odd && bounds.up_k - d < k && k < bounds.up_k + d &&
let x = down_vector.get_i32(down_offset + k); up_vector[bounds.get_up_index(k, None)] <= down_vector[bounds.get_down_index(k, None)] {
let y = add_i32(down_vector.get_i32(down_offset + k), -k); let x = down_vector[bounds.get_down_index(k, None)];
let y = add_i32(down_vector[bounds.get_down_index(k, None)], -k);
return SmsData { x, y }; return SmsData { x, y };
} }
} }
// Extend the reverse path // Extend the reverse path
for k in ((up_k - d)..=(up_k + d)).step_by(2) { for k in ((bounds.up_k - d)..=(bounds.up_k + d)).step_by(2) {
let mut x; let mut x;
let mut y; let mut y;
if k == up_k + d { if k == bounds.up_k + d {
// Up // Up
x = up_vector.get_i32(up_offset + k - 1); x = up_vector[bounds.get_up_index(k, RightUp)];
} else { } else {
// Left // Left
x = up_vector.get_i32(up_offset + k + 1) - 1; x = up_vector[bounds.get_up_index(k, LeftDown)] - 1;
if k > up_k - d && up_vector.get_i32(up_offset + k - 1) < x { if k > bounds.up_k - d && up_vector[bounds.get_up_index(k, RightUp)] < x {
// Up // Up
x = up_vector.get_i32(up_offset + k - 1); x = up_vector[bounds.get_up_index(k, RightUp)];
} }
} }
y = add_i32(x, -k); y = add_i32(x, -k);
while x > lower_a && y > lower_b && data_a.codes[x - 1] == data_b.codes[y - 1] { while x > bounds.lower_a && y > bounds.lower_b && data_a.codes[x - 1] == data_b.codes[y - 1] {
x -= 1; x -= 1;
y -= 1; y -= 1;
} }
up_vector.set_i32(up_offset + k, x); up_vector[bounds.get_up_index(k, None)] = x;
// Overlap ? // Overlap ?
if !odd_delta && down_k - d <= k && k <= down_k + d && up_vector.get_i32(up_offset + k) <= down_vector.get_i32(down_offset + k) { if !bounds.is_delta_odd && bounds.down_k - d <= k && k <= bounds.down_k + d &&
let x = down_vector.get_i32(down_offset + k); up_vector[bounds.get_up_index(k, None)] <= down_vector[bounds.get_down_index(k, None)] {
let y = add_i32(down_vector.get_i32(down_offset + k), -k); let x = down_vector[bounds.get_down_index(k, None)];
let y = add_i32(down_vector[bounds.get_down_index(k, None)], -k);
return SmsData { x, y }; return SmsData { x, y };
} }
} }

View File

@ -1,13 +1,21 @@
use std::fs::File;
use std::io;
use std::io::{BufRead, BufReader};
use crate::diff::diff; use crate::diff::diff;
mod matrix; mod matrix;
mod lcs; mod lcs;
mod diff; mod diff;
fn main() { fn main() -> io::Result<()> {
let a = "abcabba\nlkajsdfasdf\nasdfasdfasdf\nlasjkdf"; let file = File::open("test-data/config.jsonc")?;
let b = "abcabba\ncbabasdfasdf\nlasjkdf\nope"; let mut reader_a = BufReader::new(file);
let result = diff(a, b); let file = File::open("test-data/config-old.jsonc")?;
let mut reader_b = BufReader::new(file);
let result = diff(&mut reader_a, &mut reader_b);
dbg!(result); dbg!(result);
Ok(())
} }

View File

@ -0,0 +1,65 @@
{
"position": "top",
"modules-left": ["sway/workspaces"],
"modules-right": ["network", "pulseaudio", "clock"],
// Modules configuration
"sway/workspaces": {
"disable-scroll": true,
"all-outputs": true,
"format": "{icon}",
"persistent_workspaces": {
"1": [],
"2": [],
"3": [],
"4": [],
"5": [],
"6": [],
"7": [],
"8": [],
"9": [],
"10": []
},
"format-icons": {
"default": "",
"urgent": "",
"focused": ""
}
},
"custom/spotify": {
"format": "<span foreground='#abc123'> </span><span font='FireCodeMono Nerd Font Mono weight=325 Italic'>{}</span>",
"interval": 1,
"exec-if": "pgrep spotify",
"on-click": "playerctl -p spotify play-pause",
"on-scroll-up": "playerctl -p spotify previous",
"on-scroll-down": "playerctl -p spotify next",
"tooltip": false,
"escape": true,
"MAX-LENGTH": 60,
"exec": "/home/loki/bin/spotify.sh"
},
"clock": {
"format": "{:%a %d %H:%M} <span foreground='#123abc'></span>",
"tooltip-format": "<big>{:%Y %B}</big>\n<tt><small>{calendar}</small></tt>"
},
"network": {
"forletmat-disconnected": "Disconnected <span class='#abc123'></span>",
"format-ethernet": "{ipaddr} <span foreground='#123abc'></span>"
},
"pulseaudio": {
"format": "{volume}% <span foreground='#123abc'>{icon}</span>",
"format-bluetooth": "{volume}% <span foreground='#abc123'>{icon}</span>",
"format-muted": "",
"format-icons": {
"headphone": "",
"hands-free": "",
"headset": "",
"phone": "",
"portable": "",
"car": "",
"default": ["", ""]
},
"scroll-step": 1,
"on-click": "pavucontrol",
"ignored-sinks": ["Easy Effects Sink"]
}
}

65
test-data/config.jsonc Normal file
View File

@ -0,0 +1,65 @@
{
"position": "top",
"modules-left": ["sway/workspaces"],
"modules-right": ["network", "pulseaudio", "clock"],
// Modules configuration
"sway/workspaces": {
"disable-scroll": true,
"all-outputs": true,
"format": "{icon}",
"persistent_workspaces": {
"1": [],
"2": [],
"3": [],
"4": [],
"5": [],
"6": [],
"7": [],
"8": [],
"9": [],
"10": []
},
"format-icons": {
"default": "",
"urgent": "",
"focused": ""
}
},
"custom/spotify": {
"format": "<span foreground='#a4b9ef'> </span><span font='FireCodeMono Nerd Font Mono weight=325 Italic'>{}</span>",
"interval": 1,
"exec-if": "pgrep spotify",
"on-click": "playerctl -p spotify play-pause",
"on-scroll-up": "playerctl -p spotify previous",
"on-scroll-down": "playerctl -p spotify next",
"tooltip": false,
"escape": true,
"MAX-LENGTH": 60,
"exec": "/home/loki/bin/spotify.sh"
},
"clock": {
"format": "{:%a %d %H:%M} <span foreground='#a4b9ef'></span>",
"tooltip-format": "<big>{:%Y %B}</big>\n<tt><small>{calendar}</small></tt>"
},
"network": {
"format-disconnected": "Disconnected <span class='#f9c096'></span>",
"format-ethernet": "{ipaddr} <span foreground='#a4b9ef'></span>"
},
"pulseaudio": {
"format": "{volume}% <span foreground='#a4b9ef'>{icon}</span>",
"format-bluetooth": "{volume}% <span foreground='#a4b9ef'>{icon}</span>",
"format-muted": "",
"format-icons": {
"headphone": "",
"hands-free": "",
"headset": "",
"phone": "",
"portable": "",
"car": "",
"default": ["", ""]
},
"scroll-step": 1,
"on-click": "pavucontrol",
"ignored-sinks": ["Easy Effects Sink"]
}
}