com.gregkilmer.www/services/content/src/rules.ts

/**
 * The rules every change runs, in a fixed order. Pure functions: the same
 * change against the same accepted state always gets the same answer, and no
 * model decides anything.
 *
 * The rules live here in code, not in the vault, so no agent can loosen them.
 */
import { notePath, typeDef, type CheckedNote } from "@gk/schema";
import type { Author, HeldReason, StoredNote } from "./store.ts";

export interface RuleSettings {
  /** A note with at least this much body text… */
  textLossMinChars: number;
  /** …is held when an edit keeps less than this share of it. */
  textLossKeepRatio: number;
  /** An agent making more changes than this inside the window is held. */
  agentBatchLimit: number;
  agentBatchWindowMinutes: number;
  /** Note types an agent may publish without review. */
  agentAutoPublish: string[];
  /** Email addresses that may appear in notes. */
  allowedEmails: string[];
}

export const defaultRules: RuleSettings = {
  textLossMinChars: 400,
  textLossKeepRatio: 0.5,
  agentBatchLimit: 10,
  agentBatchWindowMinutes: 10,
  agentAutoPublish: ["status", "library-item"],
  allowedEmails: [],
};

export interface Proposed {
  path: string;
  name: string;
  slug: string;
  text: string;
  author: Author;
  checked: CheckedNote;
}

export interface RuleContext {
  /** The accepted version at this path, if any. */
  current: StoredNote | null;
  /** Another live note with the same file name, if any. */
  nameTakenBy: string | null;
  /** Another live note of the same type with the same slug, if any. */
  slugTakenBy: string | null;
  /** The agent's changes inside the batch window, before this one. */
  recentAgentChanges: number;
  settings: RuleSettings;
}

/**
 * Authors held to the agent rules. A sync job writes as `sync` and may do
 * nothing an agent cannot: the same contact, publish, edit and batch limits.
 */
export const isLimited = (a: Author) => a === "agent" || a === "sync";

export type Verdict = { decision: "accept" } | { decision: "hold"; reasons: HeldReason[] };

/** Strings that look like credentials. Matching is on new text only. */
const SECRET_PATTERNS: [string, RegExp][] = [
  ["a private key", /-----BEGIN [A-Z ]*PRIVATE KEY-----/],
  ["an AWS key", /\bAKIA[0-9A-Z]{16}\b/],
  ["a GitHub token", /\b(?:ghp|gho|ghs|ghu|ghr)_[A-Za-z0-9]{30,}\b|\bgithub_pat_[A-Za-z0-9_]{30,}\b/],
  ["a Slack token", /\bxox[abpors]-[A-Za-z0-9-]{10,}\b/],
  ["an API key", /\bsk-[A-Za-z0-9_-]{20,}\b/],
  ["a bearer token", /\bBearer\s+[A-Za-z0-9._~+/-]{20,}=*/],
  ["a password or key value", /\b(?:password|passwd|secret|api[_-]?key|token)\s*[:=]\s*["']?[^\s"']{8,}/i],
];
const EMAIL = /[A-Za-z0-9._%+-]+@[A-Za-z0-9.-]+\.[A-Za-z]{2,}/g;
const PHONE = /(?:\+?1[\s.-]?)?\(?\b\d{3}\)?[\s.-]\d{3}[\s.-]\d{4}\b/g;

function newMatches(re: RegExp, text: string, before: string): string[] {
  const flags = re.flags.includes("g") ? re.flags : `${re.flags}g`;
  const g = new RegExp(re.source, flags);
  const old = new Set(before.match(g) ?? []);
  return [...new Set(text.match(g) ?? [])].filter((m) => !old.has(m));
}

const bodyOf = (text: string) => text.replace(/^---\r?\n[\s\S]*?\r?\n---\r?\n?/, "").trim();

export function evaluate(p: Proposed, ctx: RuleContext): Verdict {
  const reasons: HeldReason[] = [];
  const { current, settings } = ctx;
  const live = current && !current.removed ? current : null;

  // 1. The note must be valid. Nothing else is judged on a broken note.
  if (!p.checked.ok) {
    return {
      decision: "hold",
      reasons: [
        {
          code: "invalid",
          message: p.checked.problems.map((x) => (x.field ? `${x.field}: ${x.message}` : x.message)).join("; "),
        },
      ],
    };
  }
  const data = p.checked.data;

  // 2. Settings notes live at one fixed path, so there is only ever one.
  if (typeDef(p.checked.type)?.singleton && notePath(data) !== p.path) {
    reasons.push({ code: "misplaced", message: `A ${p.checked.type} note belongs at ${notePath(data)}` });
  }

  // 3. Names and slugs must stay unambiguous.
  if (ctx.nameTakenBy) {
    reasons.push({ code: "duplicate-name", message: `Another note already uses this file name: ${ctx.nameTakenBy}` });
  }
  if (ctx.slugTakenBy) {
    reasons.push({ code: "duplicate-slug", message: `Another ${p.checked.type} already uses the address "${p.slug}": ${ctx.slugTakenBy}` });
  }

  // 4. A large loss of text in one edit.
  if (live) {
    const before = bodyOf(live.text).length;
    const after = bodyOf(p.text).length;
    if (before >= settings.textLossMinChars && after < before * settings.textLossKeepRatio) {
      reasons.push({
        code: "text-loss",
        message: `The text shrank from ${before} to ${after} characters in one edit`,
      });
    }
  }

  // 5. Anything that looks like a secret, or contact details from an agent.
  const before = live?.text ?? "";
  for (const [what, re] of SECRET_PATTERNS) {
    if (newMatches(re, p.text, before).length) reasons.push({ code: "secret", message: `The note now contains what looks like ${what}` });
  }
  const who = p.author === "sync" ? "A sync" : "An agent";
  if (isLimited(p.author)) {
    const allowed = new Set(settings.allowedEmails.map((e) => e.toLowerCase()));
    const emails = newMatches(EMAIL, p.text, before).filter((e) => !allowed.has(e.toLowerCase()));
    if (emails.length) reasons.push({ code: "contact", message: `${who} added an email address: ${emails.join(", ")}` });
    if (newMatches(PHONE, p.text, before).length) reasons.push({ code: "contact", message: `${who} added a phone number` });
  }

  // 6. Agents stay inside their lane.
  if (isLimited(p.author)) {
    // A sync may change notes it wrote itself; anything else needs agentEdits.
    if (live && live.author !== p.author && live.data.agentEdits !== true) {
      reasons.push({ code: "agent-edit", message: `${who} changed a note you wrote, which needs your review` });
    }
    const publishing = data.publish === true && !(live?.publish ?? false);
    if (publishing && !settings.agentAutoPublish.includes(p.checked.type)) {
      reasons.push({ code: "agent-publish", message: `${who} wants to publish a ${p.checked.type}, which needs your approval` });
    }
    if (ctx.recentAgentChanges >= settings.agentBatchLimit) {
      reasons.push({
        code: "batch",
        message: `More than ${settings.agentBatchLimit} ${p.author} changes in ${settings.agentBatchWindowMinutes} minutes`,
      });
    }
  }

  return reasons.length ? { decision: "hold", reasons } : { decision: "accept" };
}