ClaudeMarketing/lib/scanner.ts

import { readdirSync, readFileSync, statSync, existsSync } from "fs";
import path from "path";
import { prisma } from "./prisma";

interface ScannedFile {
  filePath: string;
  fileName: string;
  type: string;
  platform: string | null;
  format: string;
  dimensions: string | null;
  metadata: string | null;
}

/** Truncate a JSON string to maxLen while keeping it valid JSON. */
function safeJsonTruncate(json: string, maxLen: number): string {
  if (json.length <= maxLen) return json;
  try {
    const parsed = JSON.parse(json);
    // Try to produce a shorter version by re-stringifying with key limits
    const trimmed = JSON.stringify(parsed, (_key, value) => {
      if (typeof value === "string" && value.length > 200) {
        return value.slice(0, 200) + "…";
      }
      if (Array.isArray(value) && value.length > 5) {
        return value.slice(0, 5);
      }
      return value;
    });
    if (trimmed.length <= maxLen) return trimmed;
    // Still too long — return a minimal summary
    return JSON.stringify({ _truncated: true, _originalLength: json.length });
  } catch {
    return JSON.stringify({ _truncated: true, _originalLength: json.length });
  }
}

const FORMAT_TO_TYPE: Record<string, string> = {
  png: "image",
  jpg: "image",
  jpeg: "image",
  webp: "image",
  gif: "image",
  mp4: "video",
  webm: "video",
  json: "copy",
  txt: "copy",
  md: "research",
  html: "research",
};

function inferPlatform(fileName: string, filePath: string): string | null {
  const lower = (fileName + filePath).toLowerCase();
  if (lower.includes("instagram") || lower.includes("ig_") || lower.includes("ig ")) return "instagram";
  if (lower.includes("tiktok") || lower.includes("tt_") || lower.includes("tik_tok")) return "tiktok";
  if (lower.includes("nextdoor") || lower.includes("nd_")) return "nextdoor";
  return null;
}

function inferDimensions(fileName: string): string | null {
  const match = fileName.match(/(\d{3,4})x(\d{3,4})/);
  return match ? `${match[1]}x${match[2]}` : null;
}

function inferTypeFromPath(filePath: string, format: string): string {
  const lower = filePath.toLowerCase();
  if (lower.includes("/ads/")) return "image";
  if (lower.includes("/video/")) return "video";
  if (lower.includes("/copy/")) return "copy";
  if (lower.includes("/scripts/")) return "script";
  return FORMAT_TO_TYPE[format] || "research";
}

/**
 * Try to read metadata from adjacent JSON files or manifest.
 * For an image "instagram_feed_hook_a_1080x1080.png", look for:
 * - "instagram_feed_hook_a_1080x1080.json" (same name, .json ext)
 * - "ad_manifest.json" in same directory
 * For copy JSON files, read the file itself as metadata.
 */
function loadMetadata(fullPath: string, format: string): string | null {
  try {
    // For JSON files, read the content as metadata
    if (format === "json") {
      const content = readFileSync(fullPath, "utf-8");
      const parsed = JSON.parse(content);
      // Extract caption/summary if it's an array of captions
      if (Array.isArray(parsed)) {
        return JSON.stringify({ captions: parsed.slice(0, 3), totalVariations: parsed.length });
      }
      return safeJsonTruncate(content, 2000);
    }

    // For media files, look for adjacent JSON with same name
    const jsonPath = fullPath.replace(/\.[^.]+$/, ".json");
    if (existsSync(jsonPath)) {
      return safeJsonTruncate(readFileSync(jsonPath, "utf-8"), 2000);
    }

    // Look for manifest in same directory
    const dir = path.dirname(fullPath);
    const manifestPath = path.join(dir, "ad_manifest.json");
    if (existsSync(manifestPath)) {
      const manifest = JSON.parse(readFileSync(manifestPath, "utf-8"));
      const fileName = path.basename(fullPath);
      // Find this file's entry in the manifest
      if (Array.isArray(manifest)) {
        const entry = manifest.find((e: { fileName?: string; file?: string }) =>
          e.fileName === fileName || e.file === fileName
        );
        if (entry) {
          // Ensure style field is preserved in metadata
          return JSON.stringify(entry);
        }
      }
    }

    // Look for scene_plans.json for video files
    if (format === "mp4" || format === "webm") {
      const scenePlansPath = path.join(dir, "scene_plans.json");
      if (existsSync(scenePlansPath)) {
        const plans = JSON.parse(readFileSync(scenePlansPath, "utf-8"));
        const fileName = path.basename(fullPath);
        if (plans[fileName]) return JSON.stringify(plans[fileName]);
      }
    }
  } catch {
    // Metadata loading is best-effort
  }

  return null;
}

function scanDirectory(dir: string, baseDir: string): ScannedFile[] {
  const files: ScannedFile[] = [];

  try {
    const entries = readdirSync(dir);
    for (const entry of entries) {
      if (entry.startsWith(".")) continue;

      const fullPath = path.join(dir, entry);
      const stat = statSync(fullPath);

      if (stat.isDirectory()) {
        files.push(...scanDirectory(fullPath, baseDir));
      } else {
        const ext = path.extname(entry).toLowerCase().slice(1);
        if (!ext || ext === "gitkeep") continue;

        // Only ingest deliverable files — skip source/build artifacts
        const ASSET_EXTENSIONS = new Set([
          "png", "jpg", "jpeg", "webp", "gif",  // images
          "mp4", "webm",                          // videos
        ]);
        const CONTENT_EXTENSIONS = new Set([
          "json", "md", "txt",                    // copy/scripts/research
        ]);
        // Skip HTML source files, render scripts, and build tools
        if (!ASSET_EXTENSIONS.has(ext) && !CONTENT_EXTENSIONS.has(ext)) continue;
        // Skip known build/tool artifacts
        const SKIP_FILES = new Set([
          "tavily_search.mjs", "render_posters.mjs", "design_philosophy.md",
        ]);
        if (SKIP_FILES.has(entry)) continue;
        // Skip HTML source files in ads/ (they're build artifacts, not deliverables)
        const relativePath0 = path.relative(baseDir, fullPath);
        if (ext === "html" && relativePath0.includes("/ads/")) continue;

        const relativePath = path.relative(baseDir, fullPath);
        const type = inferTypeFromPath(relativePath, ext);
        const metadata = loadMetadata(fullPath, ext);

        files.push({
          filePath: relativePath,
          fileName: entry,
          type,
          platform: inferPlatform(entry, relativePath),
          format: ext,
          dimensions: inferDimensions(entry),
          metadata,
        });
      }
    }
  } catch {
    // Directory doesn't exist or can't be read
  }

  return files;
}

export async function scanOutputDirectory(
  campaignId: string,
  outputPath: string,
  pipelineRoot: string
) {
  const fullOutputPath = path.join(pipelineRoot, outputPath);
  const files = scanDirectory(fullOutputPath, pipelineRoot);

  let created = 0;

  for (const file of files) {
    const existing = await prisma.asset.findFirst({
      where: {
        campaignId,
        filePath: file.filePath,
      },
    });

    if (!existing) {
      await prisma.asset.create({
        data: {
          campaignId,
          type: file.type,
          platform: file.platform,
          format: file.format,
          filePath: file.filePath,
          fileName: file.fileName,
          dimensions: file.dimensions,
          metadata: file.metadata,
          status: "draft",
        },
      });
      created++;
    }
  }

  return { scanned: files.length, created };
}