Merge pull request 'Add fuzzy search using WFA' (#2) from wfa-fuzzy-search into main

Reviewed-on: #2
This commit is contained in:
Arthur Lu 2024-06-05 22:19:10 +00:00
commit bbfc77dc04
8 changed files with 388 additions and 5 deletions

View File

@ -52,4 +52,4 @@ input[type="radio"] {
div[draggable="true"] {
cursor: grab;
}
}

View File

@ -111,4 +111,16 @@ hr, * {
.spacer {
min-height: 1em;
}
.w3-select, select {
padding: 8px;
-moz-appearance: none;
-webkit-appearance: none;
appearance: none;
appearance: none;
background-image: url("data:image/svg+xml,%3Csvg viewBox='0 0 24 24' xmlns='http://www.w3.org/2000/svg'%3E%3Cstyle%3E:root%7Bcolor:%23fff%7D@media (prefers-color-scheme:light)%7B:root%7Bcolor:%23000%7D%7D%3C/style%3E%3Cpath fill-rule='evenodd' clip-rule='evenodd' d='M12.707 14.707a1 1 0 01-1.414 0l-5-5a1 1 0 011.414-1.414L12 12.586l4.293-4.293a1 1 0 111.414 1.414l-5 5z' fill='currentColor'/%3E%3C/svg%3E");
background-repeat: no-repeat;
background-position: right 8px top 50%;
background-size: 1em auto;
}

View File

@ -12,6 +12,7 @@
<link rel="stylesheet" href="css/form.css">
<script src="scripts/index.js" type="module"></script>
<script src="scripts/instance.js" type="module"></script>
<script src="modules/wfa.js" type="module"></script>
<style>
#instance-container > div {
border-bottom: 1px solid white;

328
modules/wfa.js Normal file
View File

@ -0,0 +1,328 @@
class WavefrontComponent {
constructor () {
this.lo = [0]; // lo for each wavefront
this.hi = [0]; // hi for each wavefront
this.W = []; // wavefront diag distance for each wavefront
this.A = []; // compact CIGAR for backtrace
}
// get value for wavefront=score, diag=k
get_val (score, k) {
if (this.W[score] !== undefined && this.W[score][k] !== undefined) {
return this.W[score][k];
}
else {
return NaN;
}
}
// set value for wavefront=score, diag=k
set_val (score, k, val) {
if (this.W[score]) {
this.W[score][k] = val;
}
else {
this.W[score] = [];
this.W[score][k] = val;
}
}
// get alignment traceback
get_traceback (score, k) {
if (this.A[score] !== undefined && this.A[score][k] !== undefined) {
return this.A[score][k];
}
else {
return undefined;
}
}
// set alignment traceback
set_traceback (score, k, traceback) {
if (this.A[score]) {
this.A[score][k] = traceback;
}
else {
this.A[score] = [];
this.A[score][k] = traceback;
}
}
// get hi for wavefront=score
get_hi (score) {
const hi = this.hi[score];
return isNaN(hi) ? 0 : hi;
}
// set hi for wavefront=score
set_hi (score, hi) {
this.hi[score] = hi;
}
// get lo for wavefront=score
get_lo (score) {
const lo = this.lo[score];
return isNaN(lo) ? 0 : lo;
}
// set lo for wavefront=score
set_lo (score, lo) {
this.lo[score] = lo;
}
// string representation of all wavefronts
toString () {
const traceback_str = ["OI", "EI", "OD", "ED", "SB", "IN", "DL", "EN"];
let s = "<";
let min_lo = Infinity;
let max_hi = -Infinity;
// get the min lo and max hi values across all wavefronts
for (let i = 0; i < this.W.length; i++) {
const lo = this.lo[i];
const hi = this.hi[i];
if (lo < min_lo) {
min_lo = lo;
}
if (hi > max_hi) {
max_hi = hi;
}
}
// print out two headers, one for wavefront and one for traceback
for (let k = min_lo; k <= max_hi; k++) {
s += FormatNumberLength(k, 2);
if (k < max_hi) {
s += "|";
}
}
s += ">\t<";
for (let k = min_lo; k <= max_hi; k++) {
s += FormatNumberLength(k, 2);
if (k < max_hi) {
s += "|";
}
}
s += ">\n";
// for each wavefront
for (let i = 0; i < this.W.length; i++) {
s += "[";
const lo = this.lo[i];
const hi = this.hi[i];
// print out the wavefront matrix
for (let k = min_lo; k <= max_hi; k++) {
if (this.W[i] !== undefined && this.W[i][k] !== undefined && !isNaN(this.W[i][k])) {
s += FormatNumberLength(this.W[i][k], 2);
}
else if (k < lo || k > hi) {
s += "--";
}
else {
s += " ";
}
if (k < max_hi) {
s += "|";
}
}
s += "]\t[";
// print out the traceback matrix
for (let k = min_lo; k <= max_hi; k++) {
if (this.A[i] !== undefined && this.A[i][k] !== undefined) {
s += traceback_str[this.A[i][k].toString()];
}
else if (k < lo || k > hi) {
s += "--";
}
else {
s += " ";
}
if (k < max_hi) {
s += "|";
}
}
s += "]\n";
}
return s;
}
}
const traceback = {
OpenIns: 0,
ExtdIns: 1,
OpenDel: 2,
ExtdDel: 3,
Sub: 4,
Ins: 5,
Del: 6,
End: 7
};
function FormatNumberLength (num, length) {
let r = "" + num;
while (r.length < length) {
r = " " + r;
}
return r;
}
function min (args) {
args.forEach((el, idx, arr) => {
arr[idx] = isNaN(el) ? Infinity : el;
});
const min = Math.min.apply(Math, args);
return min === Infinity ? NaN : min;
}
function max (args) {
args.forEach((el, idx, arr) => {
arr[idx] = isNaN(el) ? -Infinity : el;
});
const max = Math.max.apply(Math, args);
return max === -Infinity ? NaN : max;
}
function argmax (args) {
const val = max(args);
return args.indexOf(val);
}
export default function wf_align (s1, s2, penalties) {
const n = s1.length;
const m = s2.length;
const A_k = m - n;
const A_offset = m;
let score = 0;
const M = new WavefrontComponent();
M.set_val(0, 0, 0);
M.set_hi(0, 0);
M.set_lo(0, 0);
M.set_traceback(0, 0, traceback.End);
const I = new WavefrontComponent();
const D = new WavefrontComponent();
while (true) {
wf_extend(M, s1, n, s2, m, score);
if (M.get_val(score, A_k) >= A_offset) {
break;
}
score++;
wf_next(M, I, D, score, penalties);
}
return wf_backtrace(M, I, D, score, penalties, A_k, A_offset);
}
function wf_extend (M, s1, n, s2, m, score) {
const lo = M.get_lo(score);
const hi = M.get_hi(score);
for (let k = lo; k <= hi; k++) {
let v = M.get_val(score, k) - k;
let h = M.get_val(score, k);
if (isNaN(v) || isNaN(h)) {
continue;
}
while (s1[v] === s2[h]) {
M.set_val(score, k, M.get_val(score, k) + 1);
v++;
h++;
if (v > n || h > m) {
break;
}
}
}
}
function wf_next (M, I, D, score, penalties) {
const x = penalties.x;
const o = penalties.o;
const e = penalties.e;
const lo = min([M.get_lo(score - x), M.get_lo(score - o - e), I.get_lo(score - e), D.get_lo(score - e)]) - 1;
const hi = max([M.get_hi(score - x), M.get_hi(score - o - e), I.get_hi(score - e), D.get_hi(score - e)]) + 1;
M.set_hi(score, hi);
I.set_hi(score, hi);
D.set_hi(score, hi);
M.set_lo(score, lo);
I.set_lo(score, lo);
D.set_lo(score, lo);
for (let k = lo; k <= hi; k++) {
I.set_val(score, k, max([
M.get_val(score - o - e, k - 1),
I.get_val(score - e, k - 1)
]) + 1);
I.set_traceback(score, k, [traceback.OpenIns, traceback.ExtdIns][argmax([
M.get_val(score - o - e, k - 1),
I.get_val(score - e, k - 1)
])]);
D.set_val(score, k, max([
M.get_val(score - o - e, k + 1),
D.get_val(score - e, k + 1)
]));
D.set_traceback(score, k, [traceback.OpenDel, traceback.ExtdDel][argmax([
M.get_val(score - o - e, k + 1),
D.get_val(score - e, k + 1)
])]);
M.set_val(score, k, max([
M.get_val(score - x, k) + 1,
I.get_val(score, k),
D.get_val(score, k)
]));
M.set_traceback(score, k, [traceback.Sub, traceback.Ins, traceback.Del][argmax([
M.get_val(score - x, k) + 1,
I.get_val(score, k),
D.get_val(score, k)
])]);
}
}
function wf_backtrace (M, I, D, score, penalties, A_k) {
const traceback_CIGAR = ["I", "I", "D", "D", "X", "", "", ""];
const x = penalties.x;
const o = penalties.o;
const e = penalties.e;
let CIGAR_rev = ""; // reversed CIGAR
let tb_s = score; // traceback score
let tb_k = A_k; // traceback diag k
let current_traceback = M.get_traceback(tb_s, tb_k);
let done = false;
while (!done) {
CIGAR_rev += traceback_CIGAR[current_traceback];
switch (current_traceback) {
case traceback.OpenIns:
tb_s = tb_s - o - e;
tb_k = tb_k - 1;
current_traceback = M.get_traceback(tb_s, tb_k);
break;
case traceback.ExtdIns:
tb_s = tb_s - e;
tb_k = tb_k - 1;
current_traceback = I.get_traceback(tb_s, tb_k);
break;
case traceback.OpenDel:
tb_s = tb_s - o - e;
tb_k = tb_k + 1;
current_traceback = M.get_traceback(tb_s, tb_k);
break;
case traceback.ExtdDel:
tb_s = tb_s - e;
tb_k = tb_k + 1;
current_traceback = D.get_traceback(tb_s, tb_k);
break;
case traceback.Sub:
tb_s = tb_s - x;
// tb_k = tb_k;
current_traceback = M.get_traceback(tb_s, tb_k);
break;
case traceback.Ins:
// tb_s = tb_s;
// tb_k = tb_k;
current_traceback = I.get_traceback(tb_s, tb_k);
break;
case traceback.Del:
// tb_s = tb_s;
// tb_k = tb_k;
current_traceback = D.get_traceback(tb_s, tb_k);
break;
case traceback.End:
done = true;
break;
}
}
const CIGAR = Array.from(CIGAR_rev).reverse().join("");
return { CIGAR, score };
}

View File

@ -4,7 +4,7 @@
"description": "Front-end for ProxmoxAAS",
"type": "module",
"scripts": {
"lint": "html-validator --continue; stylelint --formatter verbose --fix **/*.css; DEBUG=eslint:cli-engine eslint --fix ."
"lint": "html-validator --continue; stylelint --formatter verbose --fix **/*.css; DEBUG=eslint:cli-engine eslint --fix scripts/"
},
"devDependencies": {
"eslint": "^8.43.0",

View File

@ -1,6 +1,7 @@
import { requestPVE, requestAPI, goToPage, setTitleAndHeader } from "./utils.js";
import { alert, dialog } from "./dialog.js";
import { setupClientSync } from "./clientsync.js";
import wf_align from "../modules/wfa.js";
window.addEventListener("DOMContentLoaded", init);
@ -38,6 +39,11 @@ async function getInstances () {
}
async function populateInstances () {
let searchCriteria = localStorage.getItem("search-criteria");
if (!searchCriteria) {
searchCriteria = "fuzzy";
localStorage.setItem("search-criteria", "fuzzy");
}
const searchQuery = document.querySelector("#search").value || null;
let criteria;
if (!searchQuery) {
@ -45,7 +51,7 @@ async function populateInstances () {
return (a.vmid > b.vmid) ? 1 : -1;
};
}
else {
else if (searchCriteria === "exact") {
criteria = (a, b) => {
const aInc = a.name.toLowerCase().includes(searchQuery.toLowerCase());
const bInc = b.name.toLowerCase().includes(searchQuery.toLowerCase());
@ -63,6 +69,27 @@ async function populateInstances () {
}
};
}
else if (searchCriteria === "fuzzy") {
const penalties = {
m: 0,
x: 1,
o: 1,
e: 1
};
criteria = (a, b) => {
// lower is better
const aAlign = wf_align(a.name.toLowerCase(), searchQuery.toLowerCase(), penalties);
const aScore = aAlign.score / a.name.length;
const bAlign = wf_align(b.name.toLowerCase(), searchQuery.toLowerCase(), penalties);
const bScore = bAlign.score / b.name.length;
if (aScore === bScore) {
return a.vmid > b.vmid ? 1 : -1;
}
else {
return aScore - bScore;
}
};
}
instances.sort(criteria);
const instanceContainer = document.querySelector("#instance-container");
instanceContainer.innerHTML = "";

View File

@ -12,6 +12,10 @@ function init () {
if (rate) {
document.querySelector("#sync-rate").value = rate;
}
const search = localStorage.getItem("search-criteria");
if (search) {
document.querySelector(`#search-${search}`).checked = true;
}
document.querySelector("#settings").addEventListener("submit", handleSaveSettings, false);
}
@ -20,5 +24,6 @@ function handleSaveSettings (event) {
const form = new FormData(document.querySelector("#settings"));
localStorage.setItem("sync-scheme", form.get("sync-scheme"));
localStorage.setItem("sync-rate", form.get("sync-rate"));
localStorage.setItem("search-criteria", form.get("search-criteria"));
window.location.reload();
}

View File

@ -51,10 +51,10 @@
<h3>Synchronization Settings</h3>
<fieldset>
<legend>App Sync Method</legend>
<label><input class="w3-radio" type="radio" id="sync-always" name="sync-scheme" value="always" required><span>Always Sync</span></label>
<label><input class="w3-radio" type="radio" id="sync-always" name="sync-scheme" value="always" required>Always Sync</label>
<p>App will always periodically synchronize with Proxmox. High resource usage.</p>
<label><input class="w3-radio" type="radio" id="sync-hash" name="sync-scheme" value="hash" required>Check For Sync</label>
<p>App will periodically synchronize only if there have been change. Medium resource usage.</p>
<p>App will periodically check for updates and synchronize only if needed. Medium resource usage.</p>
<label><input class="w3-radio" type="radio" id="sync-interrupt" name="sync-scheme" value="interrupt" required>Sync When Needed</label>
<p>App will react to changes and synchronize when changes are made. Low resource usage.</p>
</fieldset>
@ -65,6 +65,16 @@
</div>
</fieldset>
</div>
<div class="w3-card w3-padding">
<h3>Search Settings</h3>
<fieldset>
<legend>Instance Search Criteria</legend>
<label><input class="w3-radio" type="radio" id="search-exact" name="search-criteria" value="exact" required>Exact Match</label>
<p>Sorts by exact query match in instance name.</p>
<label><input class="w3-radio" type="radio" id="search-fuzzy" name="search-criteria" value="fuzzy" required>Fuzzy Match</label>
<p>Sorts by best matching to worst matching.</p>
</fieldset>
</div>
<div class="w3-container w3-center" id="form-actions">
<button class="w3-button w3-margin" id="save" type="submit">SAVE</button>
</div>