src/lib/agentLoop/errorSignatureBuilder.ts

/**
 * errorSignatureBuilder.ts — S697: Structured error diagnosis for self-healing
 *
 * Builds a rich ErrorSignature from raw error strings:
 * - 16 categories vs 8 flat regex (previous adaptiveSolver)
 * - Python traceback frame parsing (File "x.py", line N, in fn)
 * - JS/TS stack trace parsing (at fn (file:line:col))
 * - missingModule / missingSymbol extraction
 * - repairComplexity: trivial → simple → complex → requires_rewrite
 * - isTransient flag: timeout/network = retry; syntax = must change code
 * - Anti-loop fingerprint (djb2 hash, Safari-safe, no crypto)
 */

export type ErrorCategory =
  | "import_error"
  | "type_error"
  | "syntax_error"
  | "reference_error"
  | "assertion_fail"
  | "runtime_crash"
  | "timeout"
  | "network_fail"
  | "cors"
  | "permission"
  | "rate_limit"
  | "not_found"
  | "dependency"
  | "build_fail"
  | "quota"
  | "unknown";

export type RepairComplexity = "trivial" | "simple" | "complex" | "requires_rewrite";

export interface TracebackFrame {
  file:    string;
  line:    number;
  column?: number;
  fn?:     string;
}

export interface ErrorSignature {
  raw:              string;
  category:         ErrorCategory;
  exitCode?:        number;
  stderrLines:      string[];
  errorType?:       string;
  errorMessage?:    string;
  failedFile?:      string;
  failedLine?:      number;
  failedColumn?:    number;
  missingModule?:   string;
  missingSymbol?:   string;
  tracebackFrames:  TracebackFrame[];
  isTransient:      boolean;
  repairComplexity: RepairComplexity;
  tool:             string;
  args:             Record<string, unknown>;
  iteration:        number;
  attemptNumber:    number;
  isKnown:          boolean;
  knownFix?:        string;
  fingerprint:      string;
}

// ─── Frame parsers ────────────────────────────────────────────────────────────

function parsePythonTraceback(raw: string): TracebackFrame[] {
  const frames: TracebackFrame[] = [];
  const re = /File\s+"([^"]+)",\s+line\s+(\d+)(?:,\s+in\s+(\S+))?/g;
  let m: RegExpExecArray | null;
  while ((m = re.exec(raw)) !== null) {
    frames.push({ file: m[1], line: parseInt(m[2]), fn: m[3] });
  }
  return frames;
}

function parseJsStack(raw: string): TracebackFrame[] {
  const frames: TracebackFrame[] = [];
  const re = /at\s+(?:(\S+)\s+\()?([^\s:()]+):(\d+):(\d+)\)?/g;
  let m: RegExpExecArray | null;
  while ((m = re.exec(raw)) !== null) {
    frames.push({ fn: m[1] ?? undefined, file: m[2], line: parseInt(m[3]), column: parseInt(m[4]) });
  }
  return frames.slice(0, 10);
}

// ─── Extractors ───────────────────────────────────────────────────────────────

function extractMissingModule(raw: string): string | undefined {
  const jsM = raw.match(/(?:Cannot find module|Could not resolve|Module not found)[^'"]*['"]([^'"]+)['"]/i);
  if (jsM) return jsM[1];
  const pyM = raw.match(/(?:No module named|ModuleNotFoundError)[^'"]*['"]?([a-zA-Z0-9_.\-]+)['"]?/i);
  if (pyM) return pyM[1];
  const pyI = raw.match(/cannot import name\s+['"]([^'"]+)['"]/i);
  if (pyI) return pyI[1];
  return undefined;
}

function extractMissingSymbol(raw: string): string | undefined {
  const r1 = raw.match(/ReferenceError:\s+(\S+)\s+is not defined/i);
  if (r1) return r1[1];
  const r2 = raw.match(/TypeError:\s+(\S+)\s+is not a (?:function|constructor)/i);
  if (r2) return r2[1];
  const r3 = raw.match(/Cannot (?:read|set) properties of (?:undefined|null) \((?:reading|setting) '([^']+)'\)/i);
  if (r3) return r3[1];
  return undefined;
}

function extractFailedLine(raw: string): number | undefined {
  const m = raw.match(/(?::(\d+):\d+|at line (\d+)|line (\d+),|\((\d+)\))/);
  if (m) return parseInt(m[1] ?? m[2] ?? m[3] ?? m[4]);
  return undefined;
}

function extractFailedFile(raw: string): string | undefined {
  const m = raw.match(/(?:File "([^"]+)"|at\s+\S+\s+\(([^:)]+):\d+:\d+\)|Error in ([^\n:]+\.(?:ts|js|tsx|jsx|py)))/);
  if (m) return m[1] ?? m[2] ?? m[3];
  return undefined;
}

function djb2Fingerprint(category: ErrorCategory, msg: string, tool: string): string {
  const key = `${category}:${msg.slice(0, 80).replace(/\s+/g, " ")}:${tool}`;
  let h = 5381;
  for (let i = 0; i < key.length; i++) {
    h = (((h << 5) + h) ^ key.charCodeAt(i)) >>> 0;
  }
  return h.toString(16);
}

// ─── Main classifier ──────────────────────────────────────────────────────────

export function buildErrorSignature(
  raw:           string,
  tool:          string,
  args:          Record<string, unknown>,
  iteration:     number,
  attemptNumber: number,
): ErrorSignature {
  const lines = raw.split("\n").filter(l => l.trim().length > 0);

  let category:         ErrorCategory   = "unknown";
  let errorType:        string | undefined;
  let errorMessage:     string | undefined;
  let isTransient                        = false;
  let repairComplexity: RepairComplexity = "complex";
  let missingModule:    string | undefined;
  let missingSymbol:    string | undefined;
  const tracebackFrames: TracebackFrame[] = [];

  // ── Classification priority order ─────────────────────────────────────────
  if (/cannot find module|could not resolve|module not found|no module named|modulenotfounderror|importerror|cannot import name/i.test(raw)) {
    category = "import_error"; missingModule = extractMissingModule(raw);
    repairComplexity = missingModule ? "trivial" : "simple"; errorType = "ImportError";
  } else if (/referenceerror|is not defined/i.test(raw)) {
    category = "reference_error"; missingSymbol = extractMissingSymbol(raw);
    repairComplexity = missingSymbol ? "trivial" : "simple"; errorType = "ReferenceError";
  } else if (/typeerror|is not a function|cannot read propert|cannot set propert|is not a constructor/i.test(raw)) {
    category = "type_error"; missingSymbol = extractMissingSymbol(raw);
    repairComplexity = missingSymbol ? "simple" : "complex"; errorType = "TypeError";
  } else if (/syntaxerror|unexpected token|unterminated string|unexpected end|parse error|invalid syntax/i.test(raw)) {
    category = "syntax_error"; repairComplexity = "simple"; errorType = "SyntaxError";
  } else if (/assertionerror|assertion failed|expected.*to.*equal|test.*failed/i.test(raw)) {
    category = "assertion_fail"; repairComplexity = "requires_rewrite"; errorType = "AssertionError";
  } else if (/429|rate.?limit|too many requests|ratelimitexceeded/i.test(raw)) {
    category = "rate_limit"; isTransient = true; repairComplexity = "trivial";
  } else if (/401|403|forbidden|unauthorized|access denied|permission denied|eperm/i.test(raw)) {
    category = "permission"; repairComplexity = "simple";
  } else if (/cors|cross.?origin|access.control.allow.origin|blocked by cors/i.test(raw)) {
    category = "cors"; isTransient = true; repairComplexity = "trivial";
  } else if (/timeout|timed.?out|etimedout|operation.*timeout|abortError/i.test(raw)) {
    category = "timeout"; isTransient = true; repairComplexity = "trivial";
  } else if (/econnrefused|enotfound|failed to fetch|network error|net::err|fetch.*failed|connection.*refused/i.test(raw)) {
    category = "network_fail"; isTransient = true; repairComplexity = "trivial";
  } else if (/404|enoent|file not found|no such file|path.*does not exist/i.test(raw)) {
    category = "not_found"; repairComplexity = "simple";
  } else if (/quota|storage full|no space left|disk.*full|exceeded.*storage/i.test(raw)) {
    category = "quota"; repairComplexity = "complex";
  } else if (/peer dependency|version mismatch|incompatible.*version|dependency.*conflict/i.test(raw)) {
    category = "dependency"; repairComplexity = "simple";
  } else if (/build failed|compilation error|tsc.*error|vite.*build.*error|rollup.*error/i.test(raw)) {
    category = "build_fail"; repairComplexity = "complex";
  } else if (/exit code [^0]|exited with \d+|uncaught exception|unhandled.*rejection|segfault/i.test(raw)) {
    category = "runtime_crash"; repairComplexity = "complex";
  }

  // ── Traceback frames ───────────────────────────────────────────────────────
  if (/traceback \(most recent call last\)/i.test(raw)) {
    tracebackFrames.push(...parsePythonTraceback(raw));
  } else if (/\bat\s+(?:\S+\s+)?\([^)]+:\d+:\d+\)/.test(raw)) {
    tracebackFrames.push(...parseJsStack(raw));
  }

  // ── Error type/message from first error line ───────────────────────────────
  const firstErrLine = lines.find(l => /Error|error|Exception/i.test(l)) ?? lines[0] ?? "";
  const typeMatch = firstErrLine.match(/^([A-Z][a-zA-Z]+Error|[A-Z][a-zA-Z]+Exception):\s*(.+)/);
  if (typeMatch && !errorType) {
    errorType = typeMatch[1];
    errorMessage = typeMatch[2]?.trim().slice(0, 200);
  }

  // ── Numeric fields ─────────────────────────────────────────────────────────
  const failedLine   = extractFailedLine(raw);
  const failedFile   = extractFailedFile(raw);
  const colMatch     = raw.match(/:(\d+):(\d+)/);
  const failedColumn = colMatch ? parseInt(colMatch[2]) : undefined;

  // ── SelfLearning integration (best-effort) ─────────────────────────────────
  let isKnown = false; let knownFix: string | undefined;
  try {
    // Dynamic import to avoid circular dependency
    const sl = (globalThis as Record<string, unknown>)._selfLearningCache as { preCheck?: (t: string, a: Record<string, unknown>) => { hasKnownIssue: boolean; suggestion?: string } } | undefined;
    if (sl?.preCheck) { const pc = sl.preCheck(tool, args); if (pc.hasKnownIssue) { isKnown = true; knownFix = pc.suggestion; } }
  } catch { /* non-blocking */ }

  const fingerprint = djb2Fingerprint(category, errorMessage ?? raw.slice(0, 80), tool);

  return {
    raw:             raw.slice(0, 500),
    category,
    stderrLines:     lines.slice(0, 20),
    errorType,
    errorMessage,
    failedFile,
    failedLine,
    failedColumn,
    missingModule,
    missingSymbol,
    tracebackFrames,
    isTransient,
    repairComplexity,
    tool,
    args,
    iteration,
    attemptNumber,
    isKnown,
    knownFix,
    fingerprint,
  };
}