src/lib/pattern-engine.ts

/**
 * Pattern Matching Engine
 *
 * DSL syntax:
 *  - lowercase letters a-z: variable placeholder. Same letter = same character.
 *    Different letters = independent variables (may coincidentally be equal).
 *    Matches [a-z0-9].
 *  - digits 0-9: variable placeholder (same rules as letters). Matches [a-z0-9].
 *  - `#`: identical-run marker. A maximal run of `#` matches a run of identical
 *    characters. The first `#` starts a new group, subsequent `#`s backreference it.
 *    Matches [a-z0-9] (valid Telegram username chars excluding literal `_`).
 *  - `_`: literal underscore.
 *  - `\X`: literal character X (escape for when you need a literal a-z / 0-9 / #).
 *  - `[literal]`: a literal block — everything between `[` and `]` is treated as
 *    literal text (no variable interpretation). Useful for fixed words like
 *    `[vip]ab` which matches "vip" + two variables (vipdd, vipx1, vipab).
 *  - any other character: treated as a literal (regex-escaped).
 *
 * Examples (all match the user's listed usernames):
 *  - "a_b_0"  -> ^([a-z0-9])_([a-z0-9])_([a-z0-9])$              matches e_y_3, e_e_h, a_k_v
 *  - "ab###"  -> ^([a-z0-9])([a-z0-9])([a-z0-9])\3\3$             matches av000
 *  - "###ab"  -> ^([a-z0-9])\1\1([a-z0-9])([a-z0-9])$             matches tttxy
 *  - "a###a"  -> ^([a-z0-9])([a-z0-9])\2\2\1$                     matches u000u
 *  - "a###b"  -> ^([a-z0-9])([a-z0-9])\2\2([a-z0-9])$             matches u000q, kjjjf
 *  - "aa###"  -> ^([a-z0-9])\1([a-z0-9])\3\3$                     matches aaccc
 *  - "###bb"  -> ^([a-z0-9])\1\1([a-z0-9])\4$                     matches gggbb
 *  - "[vip]ab" -> ^vip([a-z0-9])([a-z0-9])$                       matches vipdd, vipx1
 */

export interface CompiledPattern {
  regex: string;
  flags: string;
  variableCount: number;
  length: number;
}

export interface PatternCompileError {
  error: string;
}

const USERNAME_CHAR = "[a-z0-9]";

function escapeRegex(s: string): string {
  return s.replace(/[.*+?^${}()|[\]\\]/g, "\\$&");
}

/**
 * Compile a DSL expression into an anchored regex source string.
 */
export function compilePattern(expr: string): CompiledPattern {
  if (!expr || expr.length === 0) {
    throw new Error("Pattern expression cannot be empty");
  }
  if (expr.length < 3) {
    throw new Error("Pattern is too short (Telegram usernames are 5-32 chars, patterns should be at least 3)");
  }
  if (expr.length > 32) {
    throw new Error("Pattern is too long (Telegram usernames are max 32 chars)");
  }

  let regex = "";
  const varGroups = new Map<string, number>();
  let groupCount = 0;

  let i = 0;
  while (i < expr.length) {
    const ch = expr[i];

    // Escape sequence -> literal
    if (ch === "\\" && i + 1 < expr.length) {
      regex += escapeRegex(expr[i + 1]);
      i += 2;
      continue;
    }

    // Literal block [....] -> everything inside is literal text
    if (ch === "[") {
      const end = expr.indexOf("]", i + 1);
      if (end === -1) {
        throw new Error("Unclosed [ literal block — missing ]");
      }
      const literalContent = expr.slice(i + 1, end);
      if (literalContent.length === 0) {
        throw new Error("Empty [ ] literal block");
      }
      regex += escapeRegex(literalContent);
      i = end + 1;
      continue;
    }

    // Identical run of #
    if (ch === "#") {
      let runLen = 0;
      while (i < expr.length && expr[i] === "#") {
        runLen++;
        i++;
      }
      groupCount++;
      const groupNum = groupCount;
      regex += `(${USERNAME_CHAR})`;
      for (let r = 1; r < runLen; r++) {
        regex += `\\${groupNum}`;
      }
      continue;
    }

    // Variable (letter or digit)
    if (/[a-z0-9]/.test(ch)) {
      if (varGroups.has(ch)) {
        regex += `\\${varGroups.get(ch)}`;
      } else {
        groupCount++;
        varGroups.set(ch, groupCount);
        regex += `(${USERNAME_CHAR})`;
      }
      i++;
      continue;
    }

    // Literal
    regex += escapeRegex(ch);
    i++;
  }

  return {
    regex: `^${regex}$`,
    flags: "i",
    variableCount: groupCount,
    length: expr.length,
  };
}

/**
 * Build a RegExp from a compiled pattern.
 */
export function buildRegExp(compiled: CompiledPattern): RegExp {
  return new RegExp(compiled.regex, compiled.flags);
}

/**
 * Test whether a username (without @) matches a compiled pattern.
 */
export function matchesPattern(username: string, compiled: CompiledPattern): boolean {
  const re = buildRegExp(compiled);
  return re.test(username);
}

/**
 * Validate a Telegram username (basic rules).
 * - 5-32 chars
 * - a-z, 0-9, _ only
 * - must start with a letter
 * - no leading/trailing underscore
 * - no consecutive underscores
 */
export function isValidTelegramUsername(username: string): boolean {
  if (!username) return false;
  if (username.length < 5 || username.length > 32) return false;
  if (!/^[a-z]/.test(username)) return false;
  if (!/^[a-z0-9_]+$/.test(username)) return false;
  if (username.startsWith("_") || username.endsWith("_")) return false;
  if (username.includes("__")) return false;
  return true;
}

/**
 * Generate example usernames that match a pattern (for preview in the dashboard).
 * Returns up to `count` examples.
 */
export function generateExamples(expr: string, count = 5): string[] {
  let compiled: CompiledPattern;
  try {
    compiled = compilePattern(expr);
  } catch {
    return [];
  }
  const re = buildRegExp(compiled);
  const examples: string[] = [];
  const seen = new Set<string>();

  // Deterministic generation: try a set of known-matching seeds + random attempts.
  // We brute-force over the pattern structure by building candidates from the DSL.
  const candidates = generateCandidates(expr);
  for (const c of candidates) {
    if (seen.has(c)) continue;
    seen.add(c);
    if (re.test(c) && isValidTelegramUsername(c)) {
      examples.push(c);
      if (examples.length >= count) break;
    }
  }

  // Random fill if not enough.
  let attempts = 0;
  while (examples.length < count && attempts < 2000) {
    attempts++;
    const candidate = randomUsernameForPattern(expr);
    if (candidate && !seen.has(candidate) && re.test(candidate) && isValidTelegramUsername(candidate)) {
      seen.add(candidate);
      examples.push(candidate);
    }
  }

  return examples;
}

/**
 * Build a small set of deterministic candidate usernames from the DSL expression
 * by substituting concrete values for each variable / run.
 */
function generateCandidates(expr: string): string[] {
  // Identify variable slots and run slots.
  const varNames = new Set<string>();
  const runs: { start: number; len: number; varName: string }[] = [];
  let i = 0;
  const structure: { type: "var" | "run" | "literal"; value: string; runLen?: number; varName?: string }[] = [];

  while (i < expr.length) {
    const ch = expr[i];
    if (ch === "\\" && i + 1 < expr.length) {
      structure.push({ type: "literal", value: expr[i + 1] });
      i += 2;
      continue;
    }
    if (ch === "[") {
      const end = expr.indexOf("]", i + 1);
      if (end === -1) break;
      structure.push({ type: "literal", value: expr.slice(i + 1, end) });
      i = end + 1;
      continue;
    }
    if (ch === "#") {
      let runLen = 0;
      let j = i;
      while (j < expr.length && expr[j] === "#") {
        runLen++;
        j++;
      }
      const varName = `__run${runs.length}`;
      runs.push({ start: i, len: runLen, varName });
      varNames.add(varName);
      structure.push({ type: "run", value: "", runLen, varName });
      i = j;
      continue;
    }
    if (/[a-z0-9]/.test(ch)) {
      varNames.add(ch);
      structure.push({ type: "var", value: "", varName: ch });
      i++;
      continue;
    }
    structure.push({ type: "literal", value: ch });
    i++;
  }

  const vars = Array.from(varNames);
  // Letter pool for variables (avoid _ for vars since _ is literal).
  const letterPool = "abcdefghijklmnopqrstuvwxyz".split("");
  // For runs we allow letters and digits.
  const runPool = "abcdefghijklmnopqrstuvwxyz0123456789".split("");

  const results: string[] = [];

  // Generate a few combinations.
  const combos: Record<string, string>[] = [];
  const maxCombos = 8;
  const pick = <T>(arr: T[], n: number): T => arr[n % arr.length];

  for (let n = 0; n < maxCombos; n++) {
    const combo: Record<string, string> = {};
    vars.forEach((v, idx) => {
      const isRun = v.startsWith("__run");
      const pool = isRun ? runPool : letterPool;
      combo[v] = pick(pool, n + idx * 3);
    });
    combos.push(combo);
  }

  for (const combo of combos) {
    let s = "";
    for (const part of structure) {
      if (part.type === "literal") {
        s += part.value;
      } else if (part.type === "var") {
        s += combo[part.varName!];
      } else if (part.type === "run") {
        s += combo[part.varName!].repeat(part.runLen!);
      }
    }
    results.push(s);
  }

  return results;
}

function randomUsernameForPattern(expr: string): string | null {
  const structure: { type: "var" | "run" | "literal"; value: string; runLen?: number; varName?: string }[] = [];
  const varNames = new Set<string>();
  let i = 0;
  let runIdx = 0;
  while (i < expr.length) {
    const ch = expr[i];
    if (ch === "\\" && i + 1 < expr.length) {
      structure.push({ type: "literal", value: expr[i + 1] });
      i += 2;
      continue;
    }
    if (ch === "[") {
      const end = expr.indexOf("]", i + 1);
      if (end === -1) break;
      structure.push({ type: "literal", value: expr.slice(i + 1, end) });
      i = end + 1;
      continue;
    }
    if (ch === "#") {
      let runLen = 0;
      let j = i;
      while (j < expr.length && expr[j] === "#") {
        runLen++;
        j++;
      }
      const varName = `__run${runIdx++}`;
      varNames.add(varName);
      structure.push({ type: "run", value: "", runLen, varName });
      i = j;
      continue;
    }
    if (/[a-z0-9]/.test(ch)) {
      varNames.add(ch);
      structure.push({ type: "var", value: "", varName: ch });
      i++;
      continue;
    }
    structure.push({ type: "literal", value: ch });
    i++;
  }

  const letters = "abcdefghijklmnopqrstuvwxyz";
  const alnum = "abcdefghijklmnopqrstuvwxyz0123456789";
  const rand = <T>(arr: T[]) => arr[Math.floor(Math.random() * arr.length)];

  const combo: Record<string, string> = {};
  for (const v of varNames) {
    const pool = v.startsWith("__run") ? alnum : letters;
    combo[v] = rand(pool.split(""));
  }

  let s = "";
  for (const part of structure) {
    if (part.type === "literal") s += part.value;
    else if (part.type === "var") s += combo[part.varName!];
    else if (part.type === "run") s += combo[part.varName!].repeat(part.runLen!);
  }
  return s;
}