Freeze separate music metadata field diagnostic

tools/media-eval/src/bin/music_fields.rs

@@ -0,0 +1,216 @@
+//! Known-pool hypothesis: separate real metadata fields, retaining verified v1 audio scores.
+use anyhow::{ensure, Context, Result};
+use searchi_semantic::{Device, GemmaEmbedder};
+use serde_json::{json, Value};
+use sha2::{Digest, Sha256};
+use std::{fs, io::Write, path::PathBuf, time::Instant};
+
+const BASELINE: &str = "f985ba639a14b8c0288b8fa7812d580f179f31c8cad0c21d67531c8a4fddc5b6";
+const POLICY: &str = "metadata-fields-v2:artist-title-album-genre-composer:2048bytes-total:neutral-file-title:max-real-evidence:no-cutoff";
+fn digest(bytes: &[u8]) -> String {
+    format!("{:x}", Sha256::digest(bytes))
+}
+fn read_baseline(bytes: &[u8]) -> Result<Value> {
+    ensure!(
+        digest(bytes) == BASELINE,
+        "Expected exact preserved v1 report"
+    );
+    let value: Value = serde_json::from_slice(bytes)?;
+    ensure!(
+        value["worker"]["model"] == "google/embeddinggemma-2" && value["worker"]["device"] == "cpu",
+        "Unexpected baseline runtime"
+    );
+    Ok(value)
+}
+fn fields(tags: &Value) -> Vec<Value> {
+    let mut remaining = 2048usize;
+    let mut result = Vec::new();
+    for field in ["artist", "title", "album", "genre", "composer"] {
+        if let Some(value) = tags[field].as_str().filter(|s| !s.trim().is_empty()) {
+            let prefix = format!("{field}: ");
+            let mut end = value.len().min(remaining.saturating_sub(prefix.len()));
+            while !value.is_char_boundary(end) {
+                end -= 1;
+            }
+            if end == 0 {
+                break;
+            }
+            let text = format!("{prefix}{}", &value[..end]);
+            remaining -= text.len();
+            result.push(json!({"field":field,"text":text}));
+        }
+    }
+    result
+}
+fn write_new(path: &PathBuf, value: &Value) -> Result<()> {
+    let mut file = fs::OpenOptions::new()
+        .write(true)
+        .create_new(true)
+        .open(path)?;
+    file.write_all(&serde_json::to_vec_pretty(value)?)?;
+    file.sync_all()?;
+    Ok(())
+}
+fn cosine(a: &[f32], b: &[f32]) -> Result<f64> {
+    ensure!(a.len() == 768 && b.len() == 768, "Expected 768 dimensions");
+    ensure!(a.iter().chain(b).all(|v| v.is_finite()), "Nonfinite vector");
+    Ok(a.iter()
+        .zip(b)
+        .map(|(a, b)| f64::from(*a) * f64::from(*b))
+        .sum())
+}
+fn main() -> Result<()> {
+    let args: Vec<_> = std::env::args_os().skip(1).collect();
+    ensure!(
+        args.len() == 3 || args.len() == 7,
+        "prepare BASELINE PLAN or run CORPUS BASELINE PLAN MODEL WORKER OUTPUT"
+    );
+    let preparing = args[0] == "prepare";
+    ensure!(
+        (preparing && args.len() == 3) || (args[0] == "run" && args.len() == 7),
+        "Invalid arguments"
+    );
+    let baseline_path = if preparing { &args[1] } else { &args[2] };
+    let baseline = read_baseline(&fs::read(baseline_path)?)?;
+    let assets:Vec<_>=baseline["plan"]["assets"].as_array().context("Assets missing")?.iter().map(|asset|json!({"id":asset["id"],"path":asset["path"],"sha256":asset["sha256"],"fields":fields(&asset["all_actual_tags"])})).collect();
+    let plan = json!({"policy":POLICY,"baseline_sha256":BASELINE,"fixture_sha256":baseline["plan"]["fixture_sha256"],"worker_sha256":baseline["worker_sha256"],"assets":assets,"queries":baseline["plan"]["queries"],"scope":"Known two-file diagnostic only. Audio maxima reused verbatim from hash-verified v1. All queries retained; no cutoff or weights. Other-file positives ineligible. Metadata never receives a timestamp and misleading tags remain unverified."});
+    let plan_path = PathBuf::from(if preparing { &args[2] } else { &args[3] });
+    if preparing {
+        return write_new(&plan_path, &plan);
+    }
+    ensure!(
+        std::env::var("SEARCHI_FORCE_CPU").as_deref() == Ok("1"),
+        "CPU required"
+    );
+    let frozen = fs::read(&plan_path)?;
+    ensure!(
+        serde_json::from_slice::<Value>(&frozen)? == plan,
+        "Frozen plan changed"
+    );
+    let root = PathBuf::from(&args[1]);
+    ensure!(
+        plan["fixture_sha256"] == digest(&fs::read(root.join("fixture.json"))?),
+        "Fixture changed"
+    );
+    for asset in &assets {
+        ensure!(
+            asset["sha256"]
+                == digest(&fs::read(
+                    root.join(asset["path"].as_str().context("Path missing")?)
+                )?),
+            "Asset changed"
+        );
+    }
+    ensure!(
+        plan["worker_sha256"] == digest(&fs::read(&args[5])?),
+        "Worker differs from retained audio baseline"
+    );
+    let output = PathBuf::from(&args[6]);
+    ensure!(!output.exists(), "Preserve existing output");
+    let mut model = GemmaEmbedder::load_with_device(&args[4], &args[5], 1, Device::Cpu)?;
+    ensure!(
+        serde_json::to_value(&model.info)? == baseline["worker"],
+        "Runtime differs from baseline"
+    );
+    let started = Instant::now();
+    let mut evidence = Vec::new();
+    for asset in &assets {
+        for field in asset["fields"].as_array().context("Fields missing")? {
+            let text = field["text"].as_str().context("Field text missing")?;
+            let mut vectors = model.embed_documents(&[text], "File metadata")?;
+            ensure!(vectors.len() == 1, "Expected one field vector");
+            evidence.push((
+                json!({"asset":asset["id"],"kind":"metadata","field":field["field"],"text":text}),
+                vectors.remove(0),
+            ));
+        }
+    }
+    let encoding_ms = started.elapsed().as_secs_f64() * 1000.0;
+    let mut rows = Vec::new();
+    for previous in baseline["rows"].as_array().context("Rows missing")? {
+        if previous.get("unsupported").is_some() {
+            rows.push(previous.clone());
+            continue;
+        }
+        let query = &previous["query"];
+        let vector = model.embed_query(query["text"].as_str().context("Query missing")?)?;
+        let mut ranked = previous["lanes"][0]["file_rankings"]
+            .as_array()
+            .context("Audio maxima missing")?
+            .clone();
+        let mut field_scores = Vec::new();
+        for (source, encoded) in &evidence {
+            let score = cosine(&vector, encoded)?;
+            field_scores.push(json!({"source":source,"score":score}));
+            let hit = ranked
+                .iter_mut()
+                .find(|hit| hit["source"]["asset"] == source["asset"])
+                .context("Missing retained file")?;
+            if score > hit["score"].as_f64().context("Score missing")? {
+                *hit = json!({"source":source,"score":score});
+            }
+        }
+        ranked.sort_by(|a, b| {
+            b["score"]
+                .as_f64()
+                .unwrap()
+                .total_cmp(&a["score"].as_f64().unwrap())
+        });
+        rows.push(json!({"query":query,"pool_quality_eligible":previous["pool_quality_eligible"],"audio_only":previous["lanes"][0]["file_rankings"],"whole_metadata_v1":previous["lanes"][1]["file_rankings"],"field_max_v2":ranked,"field_scores":field_scores}));
+    }
+    for asset in &assets {
+        ensure!(
+            asset["sha256"] == digest(&fs::read(root.join(asset["path"].as_str().unwrap()))?),
+            "Asset changed during inference"
+        );
+    }
+    write_new(
+        &output,
+        &json!({"plan":plan,"plan_sha256":digest(&frozen),"worker":model.info,"encoding_ms":encoding_ms,"metadata_inference_calls":evidence.len(),"audio_inference_calls":0,"rows":rows,"production_policy_changed":false}),
+    )
+}
+#[cfg(test)]
+mod tests {
+    use super::*;
+    #[test]
+    fn untrusted_boilerplate_and_blank_fields_are_excluded() {
+        assert!(fields(&json!({"encoder":"violin","title":" "})).is_empty());
+    }
+    #[test]
+    fn field_order_and_unicode_budget_are_stable() {
+        let values = fields(&json!({"title":"ä".repeat(2000),"artist":"Actual"}));
+        assert_eq!(values[0]["field"], "artist");
+        assert!(
+            values
+                .iter()
+                .map(|v| v["text"].as_str().unwrap().len())
+                .sum::<usize>()
+                <= 2048
+        );
+    }
+    #[test]
+    fn misleading_metadata_remains_literal() {
+        assert_eq!(
+            fields(&json!({"genre":"Wrong genre"}))[0]["text"],
+            "genre: Wrong genre"
+        );
+    }
+    #[test]
+    fn baseline_tampering_is_rejected() {
+        assert!(read_baseline(b"{}").is_err());
+    }
+    #[test]
+    fn dimension_mismatch_and_nonfinite_vectors_are_rejected() {
+        assert!(cosine(&[1.0], &[1.0]).is_err());
+        assert!(cosine(&vec![f32::NAN; 768], &vec![0.0; 768]).is_err());
+    }
+    #[test]
+    fn output_is_never_overwritten() -> Result<()> {
+        let dir = tempfile::tempdir()?;
+        let path = dir.path().join("report");
+        fs::write(&path, b"old")?;
+        assert!(write_new(&path, &json!({})).is_err());
+        assert_eq!(fs::read(path)?, b"old");
+        Ok(())
+    }
+}