import { stat, writeFile } from "node:fs/promises";
import { resolve as resolvePath } from "node:path";
import type { Session } from "neo4j-driver";

/**
 * Pure llms.txt + llms-full.txt generator for an account's published
 * KnowledgeDocument set. llmstxt.org draft format. Plugin-agnostic:
 * the caller injects its own neo4j session factory so the lib can
 * be consumed from any plugin that already has a configured driver
 * (aeo, admin's publish-site tool, future hook callers).
 */

export interface WriteLlmsTxtInput {
  accountId: string;
  siteName: string;
  siteDescription?: string;
  siteDir?: string;
}

export interface WriteLlmsTxtDeps {
  getSession: () => Session;
}

export interface WriteLlmsTxtResult {
  llmsTxt: string;
  llmsFullTxt: string;
  pageCount: number;
  skippedNoUrl: number;
  sizeBytes: { index: number; full: number };
  writtenPaths?: { index: string; full: string };
}

interface PageRow {
  title: string;
  url: string;
  summary: string;
  body: string;
}

const HEADER_RX = /^#\s/m;

function escapeMarkdown(text: string): string {
  return text.replace(/[\[\]\(\)\n\r]/g, " ").trim();
}

function summariseBody(body: string): string {
  const flat = body.replace(/\s+/g, " ").trim();
  if (flat.length <= 200) return flat;
  return flat.slice(0, 197).trimEnd() + "…";
}

export async function writeLlmsTxt(
  input: WriteLlmsTxtInput,
  deps: WriteLlmsTxtDeps,
): Promise<WriteLlmsTxtResult> {
  const session = deps.getSession();
  let rows: PageRow[] = [];
  let skippedNoUrl = 0;
  try {
    const r = await session.run(
      `MATCH (d:KnowledgeDocument {accountId: $accountId})
       RETURN d.title AS title, d.url AS url, d.summary AS summary,
              coalesce(d.body, d.abstract, '') AS body
       ORDER BY d.title ASC`,
      { accountId: input.accountId },
    );
    for (const rec of r.records) {
      const title = (rec.get("title") as string | null) ?? "";
      const url = (rec.get("url") as string | null) ?? "";
      const summary = (rec.get("summary") as string | null) ?? "";
      const body = (rec.get("body") as string | null) ?? "";
      if (!url || !title) {
        skippedNoUrl += 1;
        continue;
      }
      rows.push({
        title: escapeMarkdown(title),
        url: url.trim(),
        summary:
          escapeMarkdown(summary) || summariseBody(body) || "(no summary)",
        body: body.trim(),
      });
    }
  } finally {
    await session.close();
  }

  const indexLines: string[] = [];
  indexLines.push(`# ${escapeMarkdown(input.siteName)}`);
  indexLines.push("");
  if (input.siteDescription) {
    indexLines.push(`> ${escapeMarkdown(input.siteDescription)}`);
    indexLines.push("");
  }
  indexLines.push("## Pages");
  indexLines.push("");
  for (const row of rows) {
    indexLines.push(`- [${row.title}](${row.url}): ${row.summary}`);
  }
  indexLines.push("");
  indexLines.push(
    "<!-- generated by @maxy/aeo-llms-txt-writer; spec: https://llmstxt.org/ -->",
  );
  const llmsTxt = indexLines.join("\n");

  const fullLines: string[] = [];
  fullLines.push(`# ${escapeMarkdown(input.siteName)}`);
  fullLines.push("");
  if (input.siteDescription) {
    fullLines.push(`> ${escapeMarkdown(input.siteDescription)}`);
    fullLines.push("");
  }
  for (const row of rows) {
    fullLines.push(`# ${row.title}`);
    fullLines.push("");
    fullLines.push(`Source: ${row.url}`);
    fullLines.push("");
    if (row.body) {
      const safeBody = row.body.replace(HEADER_RX, "## ");
      fullLines.push(safeBody);
    } else {
      fullLines.push(row.summary);
    }
    fullLines.push("");
    fullLines.push("---");
    fullLines.push("");
  }
  const llmsFullTxt = fullLines.join("\n");

  const sizeBytes = {
    index: Buffer.byteLength(llmsTxt, "utf-8"),
    full: Buffer.byteLength(llmsFullTxt, "utf-8"),
  };

  process.stderr.write(
    `[aeo-llms-txt] site=${input.siteName} pages=${rows.length} skippedNoUrl=${skippedNoUrl} indexBytes=${sizeBytes.index} fullBytes=${sizeBytes.full}\n`,
  );

  let writtenPaths: { index: string; full: string } | undefined;
  if (input.siteDir) {
    const dir = resolvePath(input.siteDir);
    let dirStat;
    try {
      dirStat = await stat(dir);
    } catch (err) {
      const code = (err as NodeJS.ErrnoException)?.code;
      throw new Error(
        `siteDir does not exist (${code ?? "unknown"}): ${dir}`,
      );
    }
    if (!dirStat.isDirectory()) {
      throw new Error(`siteDir is not a directory: ${dir}`);
    }
    const indexPath = resolvePath(dir, "llms.txt");
    const fullPath = resolvePath(dir, "llms-full.txt");
    await writeFile(indexPath, llmsTxt, "utf-8");
    await writeFile(fullPath, llmsFullTxt, "utf-8");
    writtenPaths = { index: indexPath, full: fullPath };
    process.stderr.write(
      `[aeo-llms-txt] wrote site=${dir} indexBytes=${sizeBytes.index} fullBytes=${sizeBytes.full}\n`,
    );
  }

  return {
    llmsTxt,
    llmsFullTxt,
    pageCount: rows.length,
    skippedNoUrl,
    sizeBytes,
    writtenPaths,
  };
}