`scripts/kravet-dedup-sweep.js`

72 lines
#!/usr/bin/env node
/* READ-ONLY dedup sweep across the Kravet family. Paginates all KFAM vendors, collects every ACTIVE
   product's custom.manufacturer_sku -> {gid,title,status,createdAt,realPrice}. Flags any mfr_sku
   carried by >1 active product (the dup-SKU trap, e.g. GDW5772.007.0 at $410 vs $500). Writes a
   decision list with created date+time per dup so Steve can pick the keeper. NO writes/deletes. */
const https=require('https'), fs=require('fs');
const T=process.env.T, S='designer-laboratory-sandbox.myshopify.com';
const sleep=ms=>new Promise(r=>setTimeout(r,ms));
function gqlOnce(q){return new Promise((res,rej)=>{const rq=https.request(`https://${S}/admin/api/2024-10/graphql.json`,{method:'POST',headers:{'X-Shopify-Access-Token':T,'Content-Type':'application/json'}},x=>{let b='';x.on('data',d=>b+=d);x.on('end',()=>{try{res(JSON.parse(b))}catch(e){rej(b.slice(0,200))}})});rq.on('error',rej);rq.setTimeout(30000,()=>rq.destroy(new Error('timeout')));rq.write(JSON.stringify({query:q}));rq.end();});}
async function gql(q){for(let i=0;i<4;i++){try{return await gqlOnce(q);}catch(e){if(i===3)throw e;await new Promise(r=>setTimeout(r,1500*(i+1)));}}}
const KFAM=['Kravet','Kravet Couture','Kravet Design','Lee Jofa','Lee Jofa Modern','Brunschwig & Fils','Cole & Son','GP & J Baker','Clarke And Clarke','Mulberry','Threads','Baker Lifestyle','Gaston y Daniela','Andrew Martin'];
const norm=s=>(s||'').toUpperCase().replace(/\s+/g,' ').trim();
const isSample=o=>(o||[]).some(x=>String(x.value).toLowerCase()==='sample');
const fmt=iso=>{ if(!iso) return ''; const d=new Date(iso); return d.toLocaleString('en-US',{year:'numeric',month:'short',day:'numeric',hour:'numeric',minute:'2-digit'}); };

(async()=>{
  const bySku={};   // sku -> [{gid,title,status,createdAt,price,vendor}]
  for(const vendor of KFAM){
    let cursor=null, n=0;
    const vq=vendor.replace(/"/g,'\\"');
    while(true){
      const after=cursor?`, after:"${cursor}"`:'';
      const q=`{ products(first:100, query:"vendor:'${vq}' status:active"${after}){
        pageInfo{ hasNextPage endCursor }
        edges{ node{ id title createdAt
          metafield(namespace:"custom",key:"manufacturer_sku"){ value }
          variants(first:10){ edges{ node{ price selectedOptions{ value } } } } } } } }`;
      let s; try{ s=await gql(q); }catch(e){ console.log(`  ${vendor} ERR ${e}`); break; }
      const pr=s.data?.products; if(!pr) break;
      for(const e of pr.edges){ const nd=e.node; n++;
        const sku=norm(nd.metafield?.value); if(!sku) continue;
        const reals=(nd.variants?.edges||[]).map(x=>({p:parseFloat(x.node.price),s:isSample(x.node.selectedOptions)})).filter(v=>!v.s && v.p>5);
        const price=reals.length?Math.max(...reals.map(v=>v.p)):null;
        (bySku[sku]=bySku[sku]||[]).push({gid:nd.id.split('/').pop(),title:nd.title,createdAt:nd.createdAt,price,vendor});
      }
      if(!pr.pageInfo.hasNextPage) break; cursor=pr.pageInfo.endCursor; await sleep(600);
    }
    console.log(`  ${vendor.padEnd(20)} active=${n}`);
  }
  // TRUE dup = >1 active product sharing the SAME normalized title (color-level identical).
  // Pattern-SKU colorway families (same manufacturer_sku, different color titles) are NOT dups.
  const titleKey=t=>norm(t).replace(/\s*[-#]\s*\d+$/,'');   // strip trailing "-1"/"#2"
  const flat=[]; for(const [sku,arr] of Object.entries(bySku)) for(const p of arr) flat.push({...p,sku});
  // FORMAT-AWARE dup key:
  //  - colorway-unique SKU (has '.' or '/', e.g. GDW5772.007.0 / 100/1004.CS.0 / 8024109.12): a 2nd
  //    product with the SAME full SKU is a dup by definition -> group by SKU alone.
  //  - pattern-only SKU (no separator, e.g. W4132 / 8024108 = shared by a whole colorway family):
  //    require SAME title too, so colorways aren't falsely merged.
  // KEY by VENDOR too — the same full SKU (e.g. 30787.100.0) is reused across vendors (B&F vs Kravet)
  // for totally different products, so vendor must be part of the identity.
  const colorwayUnique=s=>/[.\/]/.test(s);
  const byTitle={}; for(const p of flat){ const v=norm(p.vendor); const k=colorwayUnique(p.sku)?('CW:'+v+'|'+p.sku):('PT:'+v+'|'+p.sku+' || '+titleKey(p.title)); (byTitle[k]=byTitle[k]||[]).push(p); }
  const dups=Object.entries(byTitle).filter(([k,a])=>a.length>1)
    .map(([k,a])=>[k,a.sort((x,y)=>new Date(x.createdAt)-new Date(y.createdAt))]);
  dups.sort((a,b)=>b[1].length-a[1].length);
  const out=['title_key,vendor,mfr_sku,product_id,price,created,full_title'];
  let dupProducts=0, priceDiffClusters=0;
  for(const [k,arr] of dups){ dupProducts+=arr.length;
    const prices=arr.map(p=>p.price).filter(v=>v!=null); if(new Set(prices).size>1) priceDiffClusters++;
    for(const p of arr) out.push(`"${k}",${p.vendor},${p.sku},${p.gid},${p.price??''},"${fmt(p.createdAt)}","${(p.title||'').replace(/"/g,'')}"`);
  }
  fs.writeFileSync('/tmp/kravet_dedup.csv',out.join('\n'));
  console.log(`\n=== KRAVET-FAMILY TRUE DUP SWEEP (same-title, READ-ONLY) ===`);
  console.log(`dup clusters (>1 active product, identical title): ${dups.length} | total products: ${dupProducts} | extra copies removable: ${dupProducts-dups.length}`);
  console.log(`clusters where copies have DIFFERENT prices (most urgent): ${priceDiffClusters}`);
  const perV={}; dups.forEach(([k,a])=>a.forEach(p=>perV[p.vendor]=(perV[p.vendor]||0)+1));
  console.log('dup products per vendor:'); for(const [v,c] of Object.entries(perV).sort((a,b)=>b[1]-a[1])) console.log(`  ${v.padEnd(20)} ${c}`);
  console.log('\nsample dup clusters (newest copy is usually the keeper):');
  for(const [k,arr] of dups.slice(0,12)) console.log(`  "${k.slice(0,40)}" x${arr.length} prices=[${arr.map(p=>p.price??'?').join(', ')}] ${arr[0].vendor}`);
  console.log(`\nfull list -> /tmp/kravet_dedup.csv`);
})().catch(e=>{console.error(e);process.exit(1);});