& so it begins

This commit is contained in:
user
2026-02-28 14:50:04 +02:00
commit f00381f2b6
536 changed files with 26294 additions and 0 deletions

View File

@@ -0,0 +1,7 @@
export function chunk<T>(arr: T[], size: number): T[][] {
const result = [];
for (let i = 0; i < arr.length; i += size) {
result.push(arr.slice(i, i + size));
}
return result;
}

View File

@@ -0,0 +1,264 @@
export const COUNTRIES = [
{ id: "1", name: "Afghanistan", code: "AF" },
{ id: "2", name: "Albania", code: "AL" },
{ id: "3", name: "Algeria", code: "DZ" },
{ id: "4", name: "American Samoa", code: "AS" },
{ id: "5", name: "Andorra", code: "AD" },
{ id: "6", name: "Angola", code: "AO" },
{ id: "7", name: "Anguilla", code: "AI" },
{ id: "8", name: "Antarctica", code: "AQ" },
{ id: "9", name: "Antigua and Barbuda", code: "AG" },
{ id: "10", name: "Argentina", code: "AR" },
{ id: "11", name: "Armenia", code: "AM" },
{ id: "12", name: "Aruba", code: "AW" },
{ id: "13", name: "Australia", code: "AU" },
{ id: "14", name: "Austria", code: "AT" },
{ id: "15", name: "Azerbaijan", code: "AZ" },
{ id: "16", name: "Bahamas", code: "BS" },
{ id: "17", name: "Bahrain", code: "BH" },
{ id: "18", name: "Bangladesh", code: "BD" },
{ id: "19", name: "Barbados", code: "BB" },
{ id: "20", name: "Belarus", code: "BY" },
{ id: "21", name: "Belgium", code: "BE" },
{ id: "22", name: "Belize", code: "BZ" },
{ id: "23", name: "Benin", code: "BJ" },
{ id: "24", name: "Bermuda", code: "BM" },
{ id: "25", name: "Bhutan", code: "BT" },
{ id: "26", name: "Bolivia", code: "BO" },
{ id: "27", name: "Bosnia and Herzegovina", code: "BA" },
{ id: "28", name: "Botswana", code: "BW" },
{ id: "29", name: "Bouvet Island", code: "BV" },
{ id: "30", name: "Brazil", code: "BR" },
{ id: "31", name: "British Indian Ocean Territory", code: "IO" },
{ id: "32", name: "British Virgin Islands", code: "VG" },
{ id: "33", name: "Brunei", code: "BN" },
{ id: "34", name: "Bulgaria", code: "BG" },
{ id: "35", name: "Burkina Faso", code: "BF" },
{ id: "36", name: "Burundi", code: "BI" },
{ id: "37", name: "Cambodia", code: "KH" },
{ id: "38", name: "Cameroon", code: "CM" },
{ id: "39", name: "Canada", code: "CA" },
{ id: "40", name: "Cape Verde", code: "CV" },
{ id: "41", name: "Caribbean Netherlands", code: "BQ" },
{ id: "42", name: "Cayman Islands", code: "KY" },
{ id: "43", name: "Central African Republic", code: "CF" },
{ id: "44", name: "Chad", code: "TD" },
{ id: "45", name: "Chile", code: "CL" },
{ id: "46", name: "China", code: "CN" },
{ id: "47", name: "Christmas Island", code: "CX" },
{ id: "48", name: "Cocos (Keeling) Islands", code: "CC" },
{ id: "49", name: "Colombia", code: "CO" },
{ id: "50", name: "Comoros", code: "KM" },
{ id: "51", name: "Cook Islands", code: "CK" },
{ id: "52", name: "Costa Rica", code: "CR" },
{ id: "53", name: "Croatia", code: "HR" },
{ id: "54", name: "Cuba", code: "CU" },
{ id: "55", name: "Curaçao", code: "CW" },
{ id: "56", name: "Cyprus", code: "CY" },
{ id: "57", name: "Czechia", code: "CZ" },
{ id: "58", name: "DR Congo", code: "CD" },
{ id: "59", name: "Denmark", code: "DK" },
{ id: "60", name: "Djibouti", code: "DJ" },
{ id: "61", name: "Dominica", code: "DM" },
{ id: "62", name: "Dominican Republic", code: "DO" },
{ id: "63", name: "Ecuador", code: "EC" },
{ id: "64", name: "Egypt", code: "EG" },
{ id: "65", name: "El Salvador", code: "SV" },
{ id: "66", name: "Equatorial Guinea", code: "GQ" },
{ id: "67", name: "Eritrea", code: "ER" },
{ id: "68", name: "Estonia", code: "EE" },
{ id: "69", name: "Eswatini", code: "SZ" },
{ id: "70", name: "Ethiopia", code: "ET" },
{ id: "71", name: "Falkland Islands", code: "FK" },
{ id: "72", name: "Faroe Islands", code: "FO" },
{ id: "73", name: "Fiji", code: "FJ" },
{ id: "74", name: "Finland", code: "FI" },
{ id: "75", name: "France", code: "FR" },
{ id: "76", name: "French Guiana", code: "GF" },
{ id: "77", name: "French Polynesia", code: "PF" },
{ id: "78", name: "French Southern and Antarctic Lands", code: "TF" },
{ id: "79", name: "Gabon", code: "GA" },
{ id: "80", name: "Gambia", code: "GM" },
{ id: "81", name: "Georgia", code: "GE" },
{ id: "82", name: "Germany", code: "DE" },
{ id: "83", name: "Ghana", code: "GH" },
{ id: "84", name: "Gibraltar", code: "GI" },
{ id: "85", name: "Greece", code: "GR" },
{ id: "86", name: "Greenland", code: "GL" },
{ id: "87", name: "Grenada", code: "GD" },
{ id: "88", name: "Guadeloupe", code: "GP" },
{ id: "89", name: "Guam", code: "GU" },
{ id: "90", name: "Guatemala", code: "GT" },
{ id: "91", name: "Guernsey", code: "GG" },
{ id: "92", name: "Guinea", code: "GN" },
{ id: "93", name: "Guinea-Bissau", code: "GW" },
{ id: "94", name: "Guyana", code: "GY" },
{ id: "95", name: "Haiti", code: "HT" },
{ id: "96", name: "Heard Island and McDonald Islands", code: "HM" },
{ id: "97", name: "Honduras", code: "HN" },
{ id: "98", name: "Hong Kong", code: "HK" },
{ id: "99", name: "Hungary", code: "HU" },
{ id: "100", name: "Iceland", code: "IS" },
{ id: "101", name: "India", code: "IN" },
{ id: "102", name: "Indonesia", code: "ID" },
{ id: "103", name: "Iran", code: "IR" },
{ id: "104", name: "Iraq", code: "IQ" },
{ id: "105", name: "Ireland", code: "IE" },
{ id: "106", name: "Isle of Man", code: "IM" },
{ id: "107", name: "Israel", code: "IL" },
{ id: "108", name: "Italy", code: "IT" },
{ id: "109", name: "Ivory Coast", code: "CI" },
{ id: "110", name: "Jamaica", code: "JM" },
{ id: "111", name: "Japan", code: "JP" },
{ id: "112", name: "Jersey", code: "JE" },
{ id: "113", name: "Jordan", code: "JO" },
{ id: "114", name: "Kazakhstan", code: "KZ" },
{ id: "115", name: "Kenya", code: "KE" },
{ id: "116", name: "Kiribati", code: "KI" },
{ id: "117", name: "Kosovo", code: "XK" },
{ id: "118", name: "Kuwait", code: "KW" },
{ id: "119", name: "Kyrgyzstan", code: "KG" },
{ id: "120", name: "Laos", code: "LA" },
{ id: "121", name: "Latvia", code: "LV" },
{ id: "122", name: "Lebanon", code: "LB" },
{ id: "123", name: "Lesotho", code: "LS" },
{ id: "124", name: "Liberia", code: "LR" },
{ id: "125", name: "Libya", code: "LY" },
{ id: "126", name: "Liechtenstein", code: "LI" },
{ id: "127", name: "Lithuania", code: "LT" },
{ id: "128", name: "Luxembourg", code: "LU" },
{ id: "129", name: "Macau", code: "MO" },
{ id: "130", name: "Madagascar", code: "MG" },
{ id: "131", name: "Malawi", code: "MW" },
{ id: "132", name: "Malaysia", code: "MY" },
{ id: "133", name: "Maldives", code: "MV" },
{ id: "134", name: "Mali", code: "ML" },
{ id: "135", name: "Malta", code: "MT" },
{ id: "136", name: "Marshall Islands", code: "MH" },
{ id: "137", name: "Martinique", code: "MQ" },
{ id: "138", name: "Mauritania", code: "MR" },
{ id: "139", name: "Mauritius", code: "MU" },
{ id: "140", name: "Mayotte", code: "YT" },
{ id: "141", name: "Mexico", code: "MX" },
{ id: "142", name: "Micronesia", code: "FM" },
{ id: "143", name: "Moldova", code: "MD" },
{ id: "144", name: "Monaco", code: "MC" },
{ id: "145", name: "Mongolia", code: "MN" },
{ id: "146", name: "Montenegro", code: "ME" },
{ id: "147", name: "Montserrat", code: "MS" },
{ id: "148", name: "Morocco", code: "MA" },
{ id: "149", name: "Mozambique", code: "MZ" },
{ id: "150", name: "Myanmar", code: "MM" },
{ id: "151", name: "Namibia", code: "NA" },
{ id: "152", name: "Nauru", code: "NR" },
{ id: "153", name: "Nepal", code: "NP" },
{ id: "154", name: "Netherlands", code: "NL" },
{ id: "155", name: "New Caledonia", code: "NC" },
{ id: "156", name: "New Zealand", code: "NZ" },
{ id: "157", name: "Nicaragua", code: "NI" },
{ id: "158", name: "Niger", code: "NE" },
{ id: "159", name: "Nigeria", code: "NG" },
{ id: "160", name: "Niue", code: "NU" },
{ id: "161", name: "Norfolk Island", code: "NF" },
{ id: "162", name: "North Korea", code: "KP" },
{ id: "163", name: "North Macedonia", code: "MK" },
{ id: "164", name: "Northern Mariana Islands", code: "MP" },
{ id: "165", name: "Norway", code: "NO" },
{ id: "166", name: "Oman", code: "OM" },
{ id: "167", name: "Pakistan", code: "PK" },
{ id: "168", name: "Palau", code: "PW" },
{ id: "169", name: "Palestine", code: "PS" },
{ id: "170", name: "Panama", code: "PA" },
{ id: "171", name: "Papua New Guinea", code: "PG" },
{ id: "172", name: "Paraguay", code: "PY" },
{ id: "173", name: "Peru", code: "PE" },
{ id: "174", name: "Philippines", code: "PH" },
{ id: "175", name: "Pitcairn Islands", code: "PN" },
{ id: "176", name: "Poland", code: "PL" },
{ id: "177", name: "Portugal", code: "PT" },
{ id: "178", name: "Puerto Rico", code: "PR" },
{ id: "179", name: "Qatar", code: "QA" },
{ id: "180", name: "Republic of the Congo", code: "CG" },
{ id: "181", name: "Romania", code: "RO" },
{ id: "182", name: "Russia", code: "RU" },
{ id: "183", name: "Rwanda", code: "RW" },
{ id: "184", name: "Réunion", code: "RE" },
{ id: "185", name: "Saint Barthélemy", code: "BL" },
{
id: "186",
name: "Saint Helena, Ascension and Tristan da Cunha",
code: "SH",
},
{ id: "187", name: "Saint Kitts and Nevis", code: "KN" },
{ id: "188", name: "Saint Lucia", code: "LC" },
{ id: "189", name: "Saint Martin", code: "MF" },
{ id: "190", name: "Saint Pierre and Miquelon", code: "PM" },
{ id: "191", name: "Saint Vincent and the Grenadines", code: "VC" },
{ id: "192", name: "Samoa", code: "WS" },
{ id: "193", name: "San Marino", code: "SM" },
{ id: "194", name: "Saudi Arabia", code: "SA" },
{ id: "195", name: "Senegal", code: "SN" },
{ id: "196", name: "Serbia", code: "RS" },
{ id: "197", name: "Seychelles", code: "SC" },
{ id: "198", name: "Sierra Leone", code: "SL" },
{ id: "199", name: "Singapore", code: "SG" },
{ id: "200", name: "Sint Maarten", code: "SX" },
{ id: "201", name: "Slovakia", code: "SK" },
{ id: "202", name: "Slovenia", code: "SI" },
{ id: "203", name: "Solomon Islands", code: "SB" },
{ id: "204", name: "Somalia", code: "SO" },
{ id: "205", name: "South Africa", code: "ZA" },
{ id: "206", name: "South Georgia", code: "GS" },
{ id: "207", name: "South Korea", code: "KR" },
{ id: "208", name: "South Sudan", code: "SS" },
{ id: "209", name: "Spain", code: "ES" },
{ id: "210", name: "Sri Lanka", code: "LK" },
{ id: "211", name: "Sudan", code: "SD" },
{ id: "212", name: "Suriname", code: "SR" },
{ id: "213", name: "Svalbard and Jan Mayen", code: "SJ" },
{ id: "214", name: "Sweden", code: "SE" },
{ id: "215", name: "Switzerland", code: "CH" },
{ id: "216", name: "Syria", code: "SY" },
{ id: "217", name: "São Tomé and Príncipe", code: "ST" },
{ id: "218", name: "Taiwan", code: "TW" },
{ id: "219", name: "Tajikistan", code: "TJ" },
{ id: "220", name: "Tanzania", code: "TZ" },
{ id: "221", name: "Thailand", code: "TH" },
{ id: "222", name: "Timor-Leste", code: "TL" },
{ id: "223", name: "Togo", code: "TG" },
{ id: "224", name: "Tokelau", code: "TK" },
{ id: "225", name: "Tonga", code: "TO" },
{ id: "226", name: "Trinidad and Tobago", code: "TT" },
{ id: "227", name: "Tunisia", code: "TN" },
{ id: "228", name: "Turkey", code: "TR" },
{ id: "229", name: "Turkmenistan", code: "TM" },
{ id: "230", name: "Turks and Caicos Islands", code: "TC" },
{ id: "231", name: "Tuvalu", code: "TV" },
{ id: "232", name: "Uganda", code: "UG" },
{ id: "233", name: "Ukraine", code: "UA" },
{ id: "234", name: "United Arab Emirates", code: "AE" },
{ id: "235", name: "United Kingdom", code: "GB" },
{ id: "236", name: "United States", code: "US" },
{ id: "237", name: "United States Minor Outlying Islands", code: "UM" },
{ id: "238", name: "United States Virgin Islands", code: "VI" },
{ id: "239", name: "Uruguay", code: "UY" },
{ id: "240", name: "Uzbekistan", code: "UZ" },
{ id: "241", name: "Vanuatu", code: "VU" },
{ id: "242", name: "Vatican City", code: "VA" },
{ id: "243", name: "Venezuela", code: "VE" },
{ id: "244", name: "Vietnam", code: "VN" },
{ id: "245", name: "Wallis and Futuna", code: "WF" },
{ id: "246", name: "Western Sahara", code: "EH" },
{ id: "247", name: "Yemen", code: "YE" },
{ id: "248", name: "Zambia", code: "ZM" },
{ id: "249", name: "Zimbabwe", code: "ZW" },
{ id: "250", name: "Åland Islands", code: "AX" },
];
export const COUNTRIES_SELECT = COUNTRIES.map((c) => {
return {
id: c.id,
label: `${c.code} (${c.name})`,
value: c.name.toLowerCase(),
};
});

File diff suppressed because it is too large Load Diff

View File

@@ -0,0 +1,83 @@
import type { CalendarDate } from "@internationalized/date";
export function formatDuration(ms: number): string {
const seconds = Math.floor(ms / 1000);
if (seconds < 60) return `${seconds}s`;
const minutes = Math.floor(seconds / 60);
const remainingSeconds = seconds % 60;
return `${minutes}m ${remainingSeconds}s`;
}
export function formatDateTimeFromIsoString(isoString: string): string {
try {
const date = new Date(isoString);
return new Intl.DateTimeFormat("en-US", {
dateStyle: "medium",
timeStyle: "short",
}).format(date);
} catch (e) {
return "Invalid date";
}
}
export function getJustDateString(d: Date): string {
return d.toISOString().split("T")[0];
}
export function formatDateTime(dateTimeStr: string) {
const date = new Date(dateTimeStr);
return {
time: date.toLocaleTimeString("en-US", {
hour: "2-digit",
minute: "2-digit",
hour12: false,
}),
date: date.toLocaleDateString("en-US", {
weekday: "short",
day: "2-digit",
month: "short",
}),
};
}
export function formatDate(dateStr: string) {
return new Date(dateStr).toLocaleDateString("en-US", {
weekday: "short",
day: "2-digit",
month: "short",
});
}
export function isTimestampMoreThan1MinAgo(ts: string): boolean {
const lastPingedDate = new Date(ts);
const now = new Date();
const diff = now.getTime() - lastPingedDate.getTime();
return diff > 60000;
}
export function isTimestampOlderThan(ts: string, seconds: number): boolean {
const lastPingedDate = new Date(ts);
const now = new Date();
const diff = now.getTime() - lastPingedDate.getTime();
return diff > seconds * 1000;
}
export function makeDateStringISO(ds: string): string {
if (ds.includes("T")) {
return `${ds.split("T")[0]}T00:00:00.000Z`;
}
return `${ds}T00:00:00.000Z`;
}
export function parseCalDateToDateString(v: CalendarDate) {
let month: string | number = v.month;
if (month < 10) {
month = `0${month}`;
}
let day: string | number = v.day;
if (day < 10) {
day = `0${day}`;
}
return `${v.year}-${month}-${day}`;
}

View File

@@ -0,0 +1,8 @@
export type Err = {
code: string;
message: string;
description: string;
detail: string;
actionable?: boolean;
error?: any;
};

View File

@@ -0,0 +1,5 @@
export type FlowExecCtx = {
flowId: string;
userId?: string;
sessionId?: string;
};

View File

@@ -0,0 +1,31 @@
import { argon2id, hash as argonHash, verify as argonVerify } from "argon2";
export async function hashString(target: string): Promise<string> {
const salt = Buffer.from(crypto.getRandomValues(new Uint8Array(16))).toString(
"hex",
);
const hash = await argonHash(target, {
type: argon2id,
salt: Buffer.from(salt, "hex"),
hashLength: 32,
timeCost: 3,
memoryCost: 65536,
parallelism: 1,
});
return hash;
}
export async function verifyHash({
hash,
target,
}: {
hash: string;
target: string;
}): Promise<boolean> {
try {
const isValid = await argonVerify(hash, `${target}`);
return isValid;
} catch (err) {
return false;
}
}

View File

@@ -0,0 +1,9 @@
import type { Session, User } from "@/domains/user/data";
import { FlowExecCtx } from "./flow.execution.context";
import { Env } from "hono";
export interface HonoContext extends Env {
Bindings: {
locals: { user: User; session: Session; fCtx: FlowExecCtx };
};
}

View File

@@ -0,0 +1,12 @@
import * as v from "valibot";
export const paginationModel = v.object({
cursor: v.optional(v.string()),
limit: v.pipe(v.number(), v.integer(), v.maxValue(100)),
asc: v.optional(v.boolean(), true),
totalItemCount: v.optional(v.pipe(v.number(), v.integer()), 0),
totalPages: v.pipe(v.number(), v.integer()),
page: v.pipe(v.number(), v.integer()),
});
export type PaginationModel = v.InferOutput<typeof paginationModel>;

View File

@@ -0,0 +1,40 @@
import { logger } from "@pkg/logger";
export class RateLimiter {
private requestTimestamps: number[] = [];
private readonly callsPerMinute: number;
constructor(callsPerMinute: number = 60) {
this.callsPerMinute = Math.min(callsPerMinute, 60);
}
async checkRateLimit(): Promise<void> {
const currentTime = Date.now();
const oneMinuteAgo = currentTime - 60000; // 60 seconds in milliseconds
// Remove timestamps older than 1 minute
this.requestTimestamps = this.requestTimestamps.filter(
(timestamp) => timestamp > oneMinuteAgo,
);
// If we're approaching the limit, wait until we have capacity
if (this.requestTimestamps.length >= this.callsPerMinute) {
const oldestRequest = this.requestTimestamps[0];
const waitTime = oldestRequest + 60000 - currentTime;
if (waitTime > 0) {
logger.warn(
`Rate limit approaching (${this.requestTimestamps.length} requests in last minute). Sleeping for ${waitTime}ms`,
);
await new Promise((resolve) => setTimeout(resolve, waitTime));
// After waiting, some timestamps may have expired
this.requestTimestamps = this.requestTimestamps.filter(
(timestamp) => timestamp > Date.now() - 60000,
);
}
}
// Add current request to timestamps
this.requestTimestamps.push(Date.now());
}
}

View File

@@ -0,0 +1 @@
export { getSetting, settings } from "@pkg/settings";

View File

@@ -0,0 +1,106 @@
import * as v from "valibot";
export function capitalize(input: string, firstOfAllWords?: boolean): string {
// capitalize first letter of input
if (!firstOfAllWords) {
return input.charAt(0).toUpperCase() + input.slice(1);
}
let out = "";
for (const word of input.split(" ")) {
out += word.charAt(0).toUpperCase() + word.slice(1) + " ";
}
return out.slice(0, -1);
}
export function camelToSpacedPascal(input: string): string {
let result = "";
let previousChar = "";
for (const char of input) {
if (char === char.toUpperCase() && previousChar !== " ") {
result += " ";
}
result += char;
previousChar = char;
}
return result.charAt(0).toUpperCase() + result.slice(1);
}
export function snakeToCamel(input: string): string {
if (!input) {
return input;
}
// also account for numbers and kebab-case
const splits = input.split(/[-_]/);
let result = splits[0];
for (const split of splits.slice(1)) {
result += capitalize(split, true);
}
return result ?? "";
}
export function snakeToSpacedPascal(input: string): string {
return camelToSpacedPascal(snakeToCamel(input));
}
export function spacedPascalToSnake(input: string): string {
return input.split(" ").join("_").toLowerCase();
}
export function convertDashedLowerToTitleCase(input: string): string {
return input
.split("-")
.map(
(word) =>
word.charAt(0).toUpperCase() + word.slice(1).toLowerCase(),
)
.join(" "); // Join the words with a space
}
export function encodeCursor<T>(cursor: T): string {
try {
// Convert the object to a JSON string
const jsonString = JSON.stringify(cursor);
// Convert to UTF-8 bytes, then base64
return btoa(
encodeURIComponent(jsonString).replace(/%([0-9A-F]{2})/g, (_, p1) =>
String.fromCharCode(parseInt(p1, 16)),
),
);
} catch (error) {
console.error("Error encoding cursor:", error);
throw new Error("Failed to encode cursor");
}
}
export function decodeCursor<T>(
cursor: string,
parser: v.BaseSchema<any, T, any>,
) {
try {
// Decode base64 back to UTF-8 string
const decoded = decodeURIComponent(
Array.prototype.map
.call(atob(cursor), (c) => {
return (
"%" + ("00" + c.charCodeAt(0).toString(16)).slice(-2)
);
})
.join(""),
);
// Parse back to object
const parsedData = JSON.parse(decoded);
const result = v.safeParse(parser, parsedData);
return result.success
? { success: true, data: result.output as T }
: {
success: false,
error: new Error(
result.issues.map((i) => i.message).join(", "),
),
data: undefined,
};
} catch (error) {
console.error("Error decoding cursor:", error);
return { error: new Error("Failed to decode cursor"), data: undefined };
}
}

View File

@@ -0,0 +1,555 @@
/**
* Similar to Python's difflib.SequenceMatcher
*
* A flexible class for comparing pairs of sequences of any type.
* Uses the Ratcliff-Obershelp algorithm with "gestalt pattern matching"
* to find the longest contiguous matching subsequences.
*/
export interface Match {
/** Starting position in sequence a */
a: number;
/** Starting position in sequence b */
b: number;
/** Length of the matching block */
size: number;
}
export type OpCode = "replace" | "delete" | "insert" | "equal";
export interface OpCodeTuple {
/** Operation type */
tag: OpCode;
/** Start index in sequence a */
i1: number;
/** End index in sequence a */
i2: number;
/** Start index in sequence b */
j1: number;
/** End index in sequence b */
j2: number;
}
export type JunkFunction<T> = (element: T) => boolean;
export class SequenceMatcher<T> {
private isjunk: JunkFunction<T> | null;
private a: T[];
private b: T[];
private autojunk: boolean;
// Cached data structures for sequence b
private bjunk: Set<T>;
private bpopular: Set<T>;
private b2j: Map<T, number[]>;
// Cached results
private fullbcount: Map<T, number> | null = null;
private matchingBlocks: Match[] | null = null;
private opcodes: OpCodeTuple[] | null = null;
constructor(
isjunk: JunkFunction<T> | null = null,
a: T[] = [],
b: T[] = [],
autojunk: boolean = true,
) {
this.isjunk = isjunk;
this.a = [];
this.b = [];
this.autojunk = autojunk;
this.bjunk = new Set();
this.bpopular = new Set();
this.b2j = new Map();
this.setSeqs(a, b);
}
/**
* Set both sequences to be compared
*/
setSeqs(a: T[], b: T[]): void {
this.setSeq1(a);
this.setSeq2(b);
}
/**
* Set the first sequence to be compared
*/
setSeq1(a: T[]): void {
if (a === this.a) return;
this.a = [...a];
this.matchingBlocks = null;
this.opcodes = null;
}
/**
* Set the second sequence to be compared
*/
setSeq2(b: T[]): void {
if (b === this.b) return;
this.b = [...b];
this.matchingBlocks = null;
this.opcodes = null;
this.fullbcount = null;
this.chainB();
}
/**
* Analyze sequence b and build lookup structures
*/
private chainB(): void {
const b = this.b;
this.bjunk = new Set();
this.bpopular = new Set();
this.b2j = new Map();
// Count occurrences of each element
const elementCounts = new Map<T, number>();
for (const element of b) {
elementCounts.set(element, (elementCounts.get(element) || 0) + 1);
}
// Determine junk and popular elements
const n = b.length;
const popularThreshold = Math.floor(n / 100) + 1; // > 1% of sequence
for (const [element, count] of elementCounts) {
if (this.isjunk && this.isjunk(element)) {
this.bjunk.add(element);
} else if (this.autojunk && n >= 200 && count > popularThreshold) {
this.bpopular.add(element);
}
}
// Build position mapping for non-junk, non-popular elements
for (let i = 0; i < b.length; i++) {
const element = b[i];
if (!this.bjunk.has(element) && !this.bpopular.has(element)) {
if (!this.b2j.has(element)) {
this.b2j.set(element, []);
}
this.b2j.get(element)!.push(i);
}
}
}
/**
* Find the longest matching block in a[alo:ahi] and b[blo:bhi]
*/
findLongestMatch(
alo: number = 0,
ahi: number | null = null,
blo: number = 0,
bhi: number | null = null,
): Match {
if (ahi === null) ahi = this.a.length;
if (bhi === null) bhi = this.b.length;
let besti = alo;
let bestj = blo;
let bestsize = 0;
// Find all positions where a[i] appears in b
const j2len = new Map<number, number>();
for (let i = alo; i < ahi; i++) {
const element = this.a[i];
const positions = this.b2j.get(element) || [];
const newj2len = new Map<number, number>();
for (const j of positions) {
if (j < blo) continue;
if (j >= bhi) break;
const prevLen = j2len.get(j - 1) || 0;
const k = prevLen + 1;
newj2len.set(j, k);
if (k > bestsize) {
besti = i - k + 1;
bestj = j - k + 1;
bestsize = k;
}
}
j2len.clear();
for (const [key, value] of newj2len) {
j2len.set(key, value);
}
}
// Extend match with junk elements
while (
besti > alo &&
bestj > blo &&
!this.isBJunk(this.b[bestj - 1]) &&
this.elementsEqual(this.a[besti - 1], this.b[bestj - 1])
) {
besti--;
bestj--;
bestsize++;
}
while (
besti + bestsize < ahi &&
bestj + bestsize < bhi &&
!this.isBJunk(this.b[bestj + bestsize]) &&
this.elementsEqual(this.a[besti + bestsize], this.b[bestj + bestsize])
) {
bestsize++;
}
// Extend match with junk elements at the beginning
while (besti > alo && bestj > blo && this.isBJunk(this.b[bestj - 1])) {
besti--;
bestj--;
bestsize++;
}
// Extend match with junk elements at the end
while (
besti + bestsize < ahi &&
bestj + bestsize < bhi &&
this.isBJunk(this.b[bestj + bestsize])
) {
bestsize++;
}
return { a: besti, b: bestj, size: bestsize };
}
/**
* Return list of non-overlapping matching blocks
*/
getMatchingBlocks(): Match[] {
if (this.matchingBlocks !== null) {
return this.matchingBlocks;
}
const matches: Match[] = [];
this.getMatchingBlocksRecursive(
0,
this.a.length,
0,
this.b.length,
matches,
);
// Add sentinel
matches.push({ a: this.a.length, b: this.b.length, size: 0 });
this.matchingBlocks = matches;
return matches;
}
/**
* Recursively find matching blocks
*/
private getMatchingBlocksRecursive(
alo: number,
ahi: number,
blo: number,
bhi: number,
matches: Match[],
): void {
const match = this.findLongestMatch(alo, ahi, blo, bhi);
if (match.size > 0) {
// Recurse on the pieces before and after the match
if (alo < match.a && blo < match.b) {
this.getMatchingBlocksRecursive(
alo,
match.a,
blo,
match.b,
matches,
);
}
matches.push(match);
if (match.a + match.size < ahi && match.b + match.size < bhi) {
this.getMatchingBlocksRecursive(
match.a + match.size,
ahi,
match.b + match.size,
bhi,
matches,
);
}
}
}
/**
* Return list of 5-tuples describing how to turn a into b
*/
getOpcodes(): OpCodeTuple[] {
if (this.opcodes !== null) {
return this.opcodes;
}
let i = 0;
let j = 0;
const opcodes: OpCodeTuple[] = [];
for (const match of this.getMatchingBlocks()) {
let tag: OpCode = "equal";
if (i < match.a && j < match.b) {
tag = "replace";
} else if (i < match.a) {
tag = "delete";
} else if (j < match.b) {
tag = "insert";
}
if (tag !== "equal") {
opcodes.push({
tag,
i1: i,
i2: match.a,
j1: j,
j2: match.b,
});
}
i = match.a + match.size;
j = match.b + match.size;
// Don't add the sentinel match
if (match.size > 0) {
opcodes.push({
tag: "equal",
i1: match.a,
i2: i,
j1: match.b,
j2: j,
});
}
}
this.opcodes = opcodes;
return opcodes;
}
/**
* Return a measure of sequences' similarity (0.0-1.0)
*/
ratio(): number {
const matches = this.getMatchingBlocks()
.slice(0, -1) // Exclude sentinel
.reduce((sum, match) => sum + match.size, 0);
const total = this.a.length + this.b.length;
return total === 0 ? 1.0 : (2.0 * matches) / total;
}
/**
* Return an upper bound on ratio() relatively quickly
*/
quickRatio(): number {
if (this.fullbcount === null) {
this.fullbcount = new Map();
for (const element of this.b) {
this.fullbcount.set(
element,
(this.fullbcount.get(element) || 0) + 1,
);
}
}
let matches = 0;
const tempCounts = new Map(this.fullbcount);
for (const element of this.a) {
const count = tempCounts.get(element);
if (count && count > 0) {
matches++;
tempCounts.set(element, count - 1);
}
}
const total = this.a.length + this.b.length;
return total === 0 ? 1.0 : (2.0 * matches) / total;
}
/**
* Return an upper bound on ratio() very quickly
*/
realQuickRatio(): number {
const total = this.a.length + this.b.length;
return total === 0
? 1.0
: (2.0 * Math.min(this.a.length, this.b.length)) / total;
}
/**
* Check if element is junk in sequence b
*/
private isBJunk(element: T): boolean {
return this.bjunk.has(element);
}
/**
* Check if two elements are equal
*/
private elementsEqual(a: T, b: T): boolean {
return a === b;
}
}
/**
* Utility function to get close matches similar to Python's get_close_matches
*/
export function getCloseMatches<T>(
word: T[],
possibilities: T[][],
n: number = 3,
cutoff: number = 0.6,
): T[][] {
if (n <= 0) {
throw new Error("n must be greater than 0");
}
const matches: Array<{ sequence: T[]; ratio: number }> = [];
for (const possibility of possibilities) {
const matcher = new SequenceMatcher(null, word, possibility);
const ratio = matcher.ratio();
if (ratio >= cutoff) {
matches.push({ sequence: possibility, ratio });
}
}
// Sort by ratio (descending) and take top n
matches.sort((a, b) => b.ratio - a.ratio);
return matches.slice(0, n).map((match) => match.sequence);
}
/**
* String-specific version of SequenceMatcher for character-by-character comparison.
* This class treats strings as sequences of characters while providing a string-friendly API.
*/
export class StringSequenceMatcher {
private matcher: SequenceMatcher<string>;
constructor(
isjunk: JunkFunction<string> | null = null,
a: string = "",
b: string = "",
autojunk: boolean = true,
) {
this.matcher = new SequenceMatcher(
isjunk,
Array.from(a),
Array.from(b),
autojunk,
);
}
/**
* Set both sequences to be compared
*/
setSeqs(a: string, b: string): void {
this.matcher.setSeqs(Array.from(a), Array.from(b));
}
/**
* Set the first sequence to be compared
*/
setSeq1(a: string): void {
this.matcher.setSeq1(Array.from(a));
}
/**
* Set the second sequence to be compared
*/
setSeq2(b: string): void {
this.matcher.setSeq2(Array.from(b));
}
/**
* Find the longest matching block in a[alo:ahi] and b[blo:bhi]
*/
findLongestMatch(
alo: number = 0,
ahi: number | null = null,
blo: number = 0,
bhi: number | null = null,
): Match {
return this.matcher.findLongestMatch(alo, ahi, blo, bhi);
}
/**
* Return list of non-overlapping matching blocks
*/
getMatchingBlocks(): Match[] {
return this.matcher.getMatchingBlocks();
}
/**
* Return list of 5-tuples describing how to turn a into b
*/
getOpcodes(): OpCodeTuple[] {
return this.matcher.getOpcodes();
}
/**
* Return a measure of sequences' similarity (0.0-1.0)
*/
ratio(): number {
return this.matcher.ratio();
}
/**
* Return an upper bound on ratio() relatively quickly
*/
quickRatio(): number {
return this.matcher.quickRatio();
}
/**
* Return an upper bound on ratio() very quickly
*/
realQuickRatio(): number {
return this.matcher.realQuickRatio();
}
}
/**
* Utility function for string similarity
*/
export function getStringSimilarity(a: string, b: string): number {
const matcher = new StringSequenceMatcher(null, a, b);
return matcher.ratio();
}
/**
* Get close string matches
*/
export function getCloseStringMatches(
word: string,
possibilities: string[],
n: number = 3,
cutoff: number = 0.6,
): string[] {
if (n <= 0) {
throw new Error("n must be greater than 0");
}
const matches: Array<{ string: string; ratio: number }> = [];
for (const possibility of possibilities) {
const ratio = getStringSimilarity(word, possibility);
if (ratio >= cutoff) {
matches.push({ string: possibility, ratio });
}
}
// Sort by ratio (descending) and take top n
matches.sort((a, b) => b.ratio - a.ratio);
return matches.slice(0, n).map((match) => match.string);
}