jmartgraphix.com/server/scripts/content-cleanup.ts

/**
 * One-shot content hygiene from dogfood QA (grammar, Flickr residue, junk tags, etc.)
 *
 *   npx tsx scripts/content-cleanup.ts
 */
import { prisma } from "../src/lib/prisma.js";
import { reindexAllProjects } from "../src/lib/typesense.js";

function cleanDescription(text: string): string {
  let t = text || "";
  t = t.replace(/\bcompanies sensors\b/gi, "company's sensors");
  t = t.replace(/\b3d model\b/g, "3D model");
  t = t.replace(/\b3d models\b/g, "3D models");
  t = t.replace(/\bgaurded\b/gi, "guarded");
  t = t.replace(/\bView\s+On\s+Black\b/gi, "");
  t = t.replace(/[ \t]+\n/g, "\n");
  t = t.replace(/\n{3,}/g, "\n\n");
  return t.trim();
}

function makeShort(description: string, existing: string | null): string | null {
  const cleaned = cleanDescription(existing || description || "");
  if (!cleaned) return null;
  // Prefer first sentence if reasonably short
  const sentence = cleaned.match(/^[^.!?]+[.!?]/)?.[0];
  if (sentence && sentence.length >= 40 && sentence.length <= 180) return sentence.trim();
  if (cleaned.length <= 180) return cleaned;
  return cleaned.slice(0, 177).replace(/\s+\S*$/, "") + "…";
}

function isJunkTag(name: string): boolean {
  const t = name.trim();
  if (!t) return true;
  if (/^\d+$/.test(t)) return true;
  if (/^\d+mm$/i.test(t)) return true;
  if (/nikon|canon|nikkor|sony|sigma|tamron|exif|\bd90\b|\bd700|\bd800|e5000/i.test(t))
    return true;
  if (/^after$/i.test(t) || /^effects$/i.test(t)) return true;
  if (/viewonblack|view.?on.?black/i.test(t)) return true;
  if (t.length >= 14 && !/\s/.test(t) && /[0-9]/.test(t)) return true;
  if (/369rollinghill|frenchcreekstatepark/i.test(t)) return true;
  return false;
}

const DESCRIPTION_OVERRIDES: Record<string, string> = {
  ankle:
    "3D animated anatomical study of a human ankle, modeled and rendered for motion work and produced with Modo and After Effects.",
  "agi-engineering-v-process-diagram":
    "Motion graphics process diagram explaining AGI’s Engineering V methodology for industrial automation projects—used in technical and marketing communications.",
  "wired-for-sound-animated-tv-spot":
    "Animated television spot for Wired for Sound, combining motion graphics and 2D animation for broadcast.",
  "aikido-kokikai-federation-akf-logo":
    "Video identity sequence for the Aikido Kokikai Federation (AKF) logo—brand motion for federation communications.",
};

async function main() {
  console.log("Content cleanup starting…");

  const projects = await prisma.project.findMany({
    select: {
      id: true,
      slug: true,
      title: true,
      description: true,
      shortDescription: true,
    },
  });

  let textFixed = 0;
  for (const p of projects) {
    let description = cleanDescription(p.description || "");
    let shortDescription = p.shortDescription
      ? cleanDescription(p.shortDescription)
      : null;

    if (DESCRIPTION_OVERRIDES[p.slug]) {
      description = DESCRIPTION_OVERRIDES[p.slug];
    }

    // Eagle: shortDescription was truncated mid-word
    if (p.slug === "eagle-in-flight" && shortDescription && /tru$/.test(shortDescription)) {
      shortDescription = makeShort(description, null);
    }

    // When short is empty, a full dump, or identical to full desc — derive a blurb
    if (
      !shortDescription ||
      shortDescription === description ||
      (description && shortDescription.length > 200 && shortDescription.startsWith(description.slice(0, 40)))
    ) {
      shortDescription = makeShort(description, null);
    }

    if (
      description !== (p.description || "") ||
      shortDescription !== (p.shortDescription || null)
    ) {
      await prisma.project.update({
        where: { id: p.id },
        data: { description, shortDescription },
      });
      textFixed++;
      console.log(`  text: ${p.slug}`);
    }
  }
  console.log(`Updated text on ${textFixed} projects`);

  // Rename CAD view to avoid /portfolio/cad collision with empty CAD category
  const cadView = await prisma.portfolioView.findFirst({ where: { slug: "cad" } });
  if (cadView) {
    await prisma.portfolioView.update({
      where: { id: cadView.id },
      data: { slug: "technical-cad", name: "CAD & Technical" },
    });
    console.log("Renamed portfolio view cad → technical-cad");
  }

  // Feature a few strong recent published projects if none featured
  const featuredCount = await prisma.project.count({
    where: { featured: true, visibility: "published" },
  });
  if (featuredCount === 0) {
    const candidates = await prisma.project.findMany({
      where: { visibility: "published" },
      orderBy: [{ date: { sort: "desc", nulls: "last" } }, { displayPriority: "asc" }],
      take: 6,
      select: { id: true, title: true, slug: true },
    });
    for (const c of candidates) {
      await prisma.project.update({ where: { id: c.id }, data: { featured: true } });
      console.log(`  featured: ${c.slug}`);
    }
    console.log(`Featured ${candidates.length} projects`);
  } else {
    console.log(`Leaving ${featuredCount} existing featured projects`);
  }

  // Junk tags
  const tags = await prisma.tag.findMany({ select: { id: true, name: true, slug: true } });
  let tagsRemoved = 0;
  for (const tag of tags) {
    if (!isJunkTag(tag.name) && !isJunkTag(tag.slug)) continue;
    await prisma.projectTag.deleteMany({ where: { tagId: tag.id } });
    await prisma.tag.delete({ where: { id: tag.id } });
    tagsRemoved++;
    console.log(`  removed tag: ${tag.name}`);
  }
  console.log(`Removed ${tagsRemoved} junk tags`);

  // Merge After Effects if both tags somehow remain (handled by delete)

  try {
    const n = await reindexAllProjects();
    console.log(`Typesense reindexed (${n} projects)`);
  } catch (err) {
    console.warn("Typesense reindex failed:", err);
  }

  console.log("Content cleanup done.");
  await prisma.$disconnect();
}

main().catch(async (err) => {
  console.error(err);
  await prisma.$disconnect();
  process.exit(1);
});