From 37b95b6b85b80fd0b575c954628a2c664247cae2 Mon Sep 17 00:00:00 2001
From: Claw <matthew@pq.io>
Date: Fri, 6 Mar 2026 16:22:14 +0000
Subject: [PATCH] feat: track token usage and estimated cost per filter run in
 filter_runs.json

---
 job_filter.mjs | 10 ++++++++--
 lib/filter.mjs | 29 +++++++++++++++++++++++++++--
 2 files changed, 35 insertions(+), 4 deletions(-)

diff --git a/job_filter.mjs b/job_filter.mjs
index 89f2722..c988f32 100644
--- a/job_filter.mjs
+++ b/job_filter.mjs
@@ -114,9 +114,13 @@ async function collect(state, settings) {
   // All done — download and merge all results
   console.log('\n  All batches complete. Downloading results...');
   const resultMap = {};
+  let totalCost = 0;
+  const totalUsage = { input_tokens: 0, output_tokens: 0, cache_creation_input_tokens: 0, cache_read_input_tokens: 0 };
   for (const b of batches) {
-    const results = await downloadResults(b.batchId, apiKey, b.idMap || {});
+    const { results, usage, cost } = await downloadResults(b.batchId, apiKey, b.idMap || {});
     for (const r of results) resultMap[r.jobId] = r;
+    totalCost += cost;
+    for (const [k, v] of Object.entries(usage)) totalUsage[k] = (totalUsage[k] || 0) + v;
   }
 
   const searchConfig = loadConfig(resolve(__dir, 'config/search_config.json'));
@@ -170,10 +174,12 @@ async function collect(state, settings) {
     passed,
     filtered,
     errors,
+    cost_usd: Math.round(totalCost * 100) / 100,
+    usage: totalUsage,
   });
   writeFileSync(runsPath, JSON.stringify(runs, null, 2));
 
-  const summary = `✅ Filter complete — ${passed} passed, ${filtered} filtered, ${errors} errors`;
+  const summary = `✅ Filter complete — ${passed} passed, ${filtered} filtered, ${errors} errors (est. cost: $${totalCost.toFixed(2)})`;
   console.log(`\n${summary}`);
 
   await sendTelegram(settings,
diff --git a/lib/filter.mjs b/lib/filter.mjs
index 856b721..e9e3f3f 100644
--- a/lib/filter.mjs
+++ b/lib/filter.mjs
@@ -146,6 +146,14 @@ export async function checkBatch(batchId, apiKey) {
 /**
  * Download and parse batch results. Returns array of { jobId, score, reason, error }
  */
+// Sonnet batch API pricing (per million tokens)
+const PRICING = {
+  input:          1.50,
+  output:         7.50,
+  cache_write:    1.875,
+  cache_read:     0.15,
+};
+
 export async function downloadResults(batchId, apiKey, idMap = {}) {
   const res = await fetch(`${BATCH_API}/${batchId}/results`, {
     headers: apiHeaders(apiKey),
@@ -156,13 +164,22 @@ export async function downloadResults(batchId, apiKey, idMap = {}) {
   const text = await res.text();
   const lines = text.trim().split('\n').filter(Boolean);
   const results = [];
+  const usage = { input_tokens: 0, output_tokens: 0, cache_creation_input_tokens: 0, cache_read_input_tokens: 0 };
 
   for (const line of lines) {
     try {
       const entry = JSON.parse(line);
-      // Resolve truncated custom_id back to original job ID
       const jobId = idMap[entry.custom_id] || entry.custom_id;
 
+      // Accumulate token usage
+      const u = entry.result?.message?.usage;
+      if (u) {
+        usage.input_tokens += u.input_tokens || 0;
+        usage.output_tokens += u.output_tokens || 0;
+        usage.cache_creation_input_tokens += u.cache_creation_input_tokens || 0;
+        usage.cache_read_input_tokens += u.cache_read_input_tokens || 0;
+      }
+
       if (entry.result?.type === 'succeeded') {
         const content = entry.result.message?.content?.[0]?.text || '';
         try {
@@ -180,5 +197,13 @@ export async function downloadResults(batchId, apiKey, idMap = {}) {
     }
   }
 
-  return results;
+  // Calculate estimated cost
+  const cost = (
+    (usage.input_tokens * PRICING.input) +
+    (usage.output_tokens * PRICING.output) +
+    (usage.cache_creation_input_tokens * PRICING.cache_write) +
+    (usage.cache_read_input_tokens * PRICING.cache_read)
+  ) / 1_000_000;
+
+  return { results, usage, cost: Math.round(cost * 100) / 100 };
 }