auditmysite 1.7.3

//! Core Web Vitals extraction via CDP
//!
//! Collects LCP, FCP, CLS, TBT, TTFB and other performance metrics.

use chromiumoxide::cdp::browser_protocol::performance::GetMetricsParams;
use chromiumoxide::Page;
use serde::{Deserialize, Serialize};
use tracing::{debug, info, warn};

use crate::error::{AuditError, Result};

/// Source element attribution for a single CLS layout-shift entry.
#[derive(Debug, Clone, Serialize, Deserialize)]
pub struct ClsSource {
    /// Simplified node description, e.g. "DIV#banner" or "IMG"
    pub node: String,
    /// Bounding rect before the shift (may be absent on first paint)
    #[serde(skip_serializing_if = "Option::is_none")]
    pub previous_rect: Option<ShiftRect>,
    /// Bounding rect after the shift
    #[serde(skip_serializing_if = "Option::is_none")]
    pub current_rect: Option<ShiftRect>,
}

/// Axis-aligned bounding rect (viewport-relative pixels).
#[derive(Debug, Clone, Serialize, Deserialize)]
pub struct ShiftRect {
    pub top: f64,
    pub left: f64,
    pub bottom: f64,
    pub right: f64,
}

/// A single layout-shift entry with per-element attribution.
#[derive(Debug, Clone, Serialize, Deserialize)]
pub struct ClsShift {
    /// Score contribution of this individual shift
    pub value: f64,
    /// When the shift occurred (ms from navigation start)
    pub start_time_ms: f64,
    /// Elements that moved (may be empty when browser does not expose attribution)
    pub sources: Vec<ClsSource>,
}

/// Core Web Vitals and performance metrics
#[derive(Debug, Clone, Serialize, Deserialize, Default)]
pub struct WebVitals {
    /// Largest Contentful Paint (ms) - target ≤2500
    pub lcp: Option<VitalMetric>,
    /// First Contentful Paint (ms) - target ≤1800
    pub fcp: Option<VitalMetric>,
    /// Cumulative Layout Shift - target ≤0.1
    pub cls: Option<VitalMetric>,
    /// Per-shift CLS attribution (#135)
    #[serde(default, skip_serializing_if = "Vec::is_empty")]
    pub cls_attribution: Vec<ClsShift>,
    /// Time to First Byte (ms)
    pub ttfb: Option<VitalMetric>,
    /// First Input Delay (ms) - deprecated; always null in CDP/headless audits
    #[serde(skip_serializing_if = "Option::is_none")]
    pub fid: Option<VitalMetric>,
    /// Total Blocking Time (ms)
    pub tbt: Option<VitalMetric>,
    /// Speed Index — heuristic proxy: 0.35·FCP + 0.65·LCP (#137)
    pub speed_index: Option<VitalMetric>,
    /// Time to Interactive — simplified: max(domInteractive, last long-task end) (#137)
    pub tti: Option<VitalMetric>,
    /// Interaction to Next Paint — only populated when interaction events fire during audit
    #[serde(skip_serializing_if = "Option::is_none")]
    pub inp: Option<VitalMetric>,
    /// DOM Content Loaded (ms)
    pub dom_content_loaded: Option<f64>,
    /// Page Load Time (ms)
    pub load_time: Option<f64>,
    /// DOM Nodes count
    pub dom_nodes: Option<i64>,
    /// JavaScript Heap Size (bytes)
    pub js_heap_size: Option<i64>,
}

/// How a metric value was obtained.
///
/// Every auditmysite performance metric is **lab data** captured in a local
/// headless Chrome — never field/RUM data (e.g. Chrome UX Report / CrUX). This
/// flag distinguishes a direct headless measurement from a value that was
/// derived/estimated from other lab signals, so that estimated metrics such as
/// INP, TTI and Speed Index cannot be mistaken for real field measurements.
#[derive(Debug, Clone, Copy, PartialEq, Eq, Serialize, Deserialize, Default)]
#[serde(rename_all = "snake_case")]
pub enum MeasurementContext {
    /// Directly measured in the local headless browser via CDP / PerformanceObserver.
    #[default]
    LabHeadless,
    /// Derived or estimated from other lab signals (heuristic proxy), still local lab.
    EstimatedLab,
    /// Directly measured, but under the Lighthouse mobile throttle (slow 4G +
    /// 4x CPU). The value is intentionally more conservative than an
    /// unthrottled load — a heavy site legitimately shows a double-digit LCP.
    LabThrottledMobile,
}

/// Individual vital metric with value and rating
#[derive(Debug, Clone, Serialize, Deserialize)]
pub struct VitalMetric {
    /// The metric value
    pub value: f64,
    /// Rating: "good", "needs-improvement", "poor"
    pub rating: String,
    /// Target threshold for "good"
    pub target: f64,
    /// Provenance of the value: direct headless measurement vs. estimated lab.
    /// Never field data. Defaults to `lab_headless`.
    #[serde(default)]
    pub measurement: MeasurementContext,
}

impl VitalMetric {
    pub fn new(value: f64, good_threshold: f64, poor_threshold: f64) -> Self {
        let rating = if value <= good_threshold {
            "good"
        } else if value <= poor_threshold {
            "needs-improvement"
        } else {
            "poor"
        };

        Self {
            value,
            rating: rating.to_string(),
            target: good_threshold,
            measurement: MeasurementContext::LabHeadless,
        }
    }

    /// Mark this metric as an estimated/heuristic lab value (e.g. Speed Index,
    /// TTI, INP proxy) rather than a direct headless measurement.
    pub fn estimated(mut self) -> Self {
        self.measurement = MeasurementContext::EstimatedLab;
        self
    }

    pub fn is_good(&self) -> bool {
        self.rating == "good"
    }

    /// Whether this value was estimated rather than directly measured.
    pub fn is_estimated(&self) -> bool {
        self.measurement == MeasurementContext::EstimatedLab
    }

    /// Re-tag a directly-measured metric as captured under mobile throttling.
    /// No-op for estimated metrics — their estimation is the stronger caveat.
    pub fn into_throttled_mobile(mut self) -> Self {
        if self.measurement == MeasurementContext::LabHeadless {
            self.measurement = MeasurementContext::LabThrottledMobile;
        }
        self
    }
}

impl WebVitals {
    /// Count how many vitals pass the "good" threshold
    pub fn good_count(&self) -> usize {
        [&self.lcp, &self.fcp, &self.cls, &self.tbt]
            .iter()
            .filter(|v| v.as_ref().map(|m| m.is_good()).unwrap_or(false))
            .count()
    }

    /// Count how many vitals are available
    pub fn available_count(&self) -> usize {
        [&self.lcp, &self.fcp, &self.cls, &self.tbt]
            .iter()
            .filter(|v| v.is_some())
            .count()
    }
}

/// Inject PerformanceObserver script before navigation so LCP, TBT, and CLS
/// entries are captured from the very first paint (Lighthouse-style).
///
/// Must be called on a fresh `Page` BEFORE `browser.navigate()`. The observers
/// store results in `window.__ams_lcp`, `window.__ams_tbt`, `window.__ams_cls`.
pub async fn prepare_vitals_collection(page: &Page) -> Result<()> {
    let script = r#"
(function() {
    // Load-window boundary: LCP candidates and long tasks are only counted up
    // to (load event + 100ms grace), mirroring how Lighthouse bounds these
    // metrics to the load phase. A headless audit fires no user-input or
    // visibility-change event — the signals that normally stop LCP growth — so
    // without this boundary the observers keep absorbing post-load paints
    // (carousels, lazy hero swaps) and idle long tasks (analytics, late
    // hydration), inflating LCP/TBT/TTI. Until load fires the boundary is
    // +Infinity, so everything during loading is counted.
    if (typeof window.__ams_load_ts === 'undefined') window.__ams_load_ts = Infinity;
    window.addEventListener('load', function() {
        window.__ams_load_ts = performance.now() + 100;
    }, { once: true });

    // LCP observer — keep the latest contentful paint within the load window.
    window.__ams_lcp = window.__ams_lcp || 0;
    try {
        var _lcpObs = new PerformanceObserver(function(list) {
            var entries = list.getEntries();
            for (var i = 0; i < entries.length; i++) {
                // LCP candidates grow monotonically in size; the last one that
                // rendered within the load window is the LCP.
                if (entries[i].startTime <= window.__ams_load_ts) {
                    window.__ams_lcp = entries[i].startTime;
                }
            }
        });
        _lcpObs.observe({ type: 'largest-contentful-paint', buffered: true });
        window.__ams_lcp_obs = _lcpObs;
    } catch(e) {}

    // TBT + TTI via Long Tasks — sum blocking time; track last task end for TTI
    window.__ams_tbt = window.__ams_tbt || 0;
    window.__ams_last_task_end = window.__ams_last_task_end || 0;
    window.__ams_tbt_seen = window.__ams_tbt_seen || new Set();
    try {
        new PerformanceObserver(function(list) {
            var entries = list.getEntries();
            for (var i = 0; i < entries.length; i++) {
                var e = entries[i];
                if (window.__ams_tbt_seen.has(e.startTime)) continue;
                window.__ams_tbt_seen.add(e.startTime);
                // Only count long tasks that began within the load window; TBT
                // is the blocking time during load (FCP→TTI), not the total
                // over the whole time we keep the page open.
                if (e.startTime > window.__ams_load_ts) continue;
                if (e.duration > 50) {
                    window.__ams_tbt += e.duration - 50;
                }
                var end = e.startTime + e.duration;
                if (end > window.__ams_last_task_end) {
                    window.__ams_last_task_end = end;
                }
            }
        }).observe({ type: 'longtask', buffered: true });
    } catch(e) {}

    // CLS — accumulate total and capture per-shift attribution (#135)
    window.__ams_cls = window.__ams_cls || 0;
    window.__ams_cls_shifts = window.__ams_cls_shifts || [];
    window.__ams_cls_seen = window.__ams_cls_seen || new Set();
    try {
        new PerformanceObserver(function(list) {
            var entries = list.getEntries();
            for (var i = 0; i < entries.length; i++) {
                var e = entries[i];
                if (window.__ams_cls_seen.has(e.startTime)) continue;
                window.__ams_cls_seen.add(e.startTime);
                if (!e.hadRecentInput) {
                    window.__ams_cls += e.value;
                    var sources = [];
                    try {
                        if (e.sources) {
                            for (var j = 0; j < e.sources.length; j++) {
                                var s = e.sources[j];
                                var nodeDesc = 'unknown';
                                try {
                                    if (s.node) {
                                        nodeDesc = s.node.nodeName;
                                        if (s.node.id) nodeDesc += '#' + s.node.id;
                                        else if (s.node.className && typeof s.node.className === 'string') {
                                            var cls = s.node.className.trim().split(/\s+/)[0];
                                            if (cls) nodeDesc += '.' + cls;
                                        }
                                    }
                                } catch(_) {}
                                sources.push({
                                    node: nodeDesc,
                                    previousRect: s.previousRect ? {top: s.previousRect.top, left: s.previousRect.left, bottom: s.previousRect.bottom, right: s.previousRect.right} : null,
                                    currentRect: s.currentRect ? {top: s.currentRect.top, left: s.currentRect.left, bottom: s.currentRect.bottom, right: s.currentRect.right} : null
                                });
                            }
                        }
                    } catch(_) {}
                    window.__ams_cls_shifts.push({value: e.value, startTime: e.startTime, sources: sources});
                }
            }
        }).observe({ type: 'layout-shift', buffered: true });
    } catch(e) {}

    // INP via event-timing (#134) — captures real interaction durations
    window.__ams_inp = window.__ams_inp || 0;
    try {
        new PerformanceObserver(function(list) {
            var entries = list.getEntries();
            for (var i = 0; i < entries.length; i++) {
                var e = entries[i];
                if (e.duration > window.__ams_inp) {
                    window.__ams_inp = e.duration;
                }
            }
        }).observe({ type: 'event', durationThreshold: 16, buffered: true });
    } catch(e) {}
    // Also observe first-input for environments that don't support event-timing
    try {
        new PerformanceObserver(function(list) {
            var entries = list.getEntries();
            for (var i = 0; i < entries.length; i++) {
                var e = entries[i];
                var delay = e.processingStart - e.startTime;
                if (delay > window.__ams_inp) window.__ams_inp = delay;
            }
        }).observe({ type: 'first-input', buffered: true });
    } catch(e) {}
})();
"#;

    page.evaluate_on_new_document(script)
        .await
        .map_err(|e| AuditError::CdpError(format!("Failed to inject vitals observer: {}", e)))?;

    Ok(())
}

/// Read the pre-injected globals after navigation to collect LCP, TBT, CLS, TTI, INP.
///
/// Flushes and disconnects the LCP observer to capture any buffered entries.
async fn collect_preinjected_vitals(page: &Page) -> Result<PreinjectedVitals> {
    let js = r#"
(function() {
    // Finalize LCP — flush and disconnect observer
    var lcp = 0;
    try {
        if (window.__ams_lcp_obs) {
            var recs = window.__ams_lcp_obs.takeRecords();
            var bound = window.__ams_load_ts || Infinity;
            for (var i = 0; i < recs.length; i++) {
                // Honor the load-window boundary here too, so a buffered
                // post-load paint can't override the in-window LCP.
                if (recs[i].startTime <= bound) {
                    window.__ams_lcp = recs[i].startTime;
                }
            }
            window.__ams_lcp_obs.disconnect();
        }
        lcp = window.__ams_lcp || 0;
    } catch(e) {}

    return JSON.stringify({
        lcp: lcp,
        tbt: window.__ams_tbt || 0,
        cls: window.__ams_cls || 0,
        lastTaskEnd: window.__ams_last_task_end || 0,
        clsShifts: window.__ams_cls_shifts || [],
        inpDuration: window.__ams_inp || 0
    });
})()
"#;

    let result = page
        .evaluate(js)
        .await
        .map_err(|e| AuditError::CdpError(format!("Preinjected vitals read failed: {}", e)))?;

    let json_str = result.value().and_then(|v| v.as_str()).unwrap_or("{}");
    let parsed: serde_json::Value = serde_json::from_str(json_str).unwrap_or_default();

    let cls_shifts = parse_cls_shifts(&parsed["clsShifts"]);

    Ok(PreinjectedVitals {
        lcp: parsed["lcp"].as_f64().unwrap_or(0.0),
        tbt: parsed["tbt"].as_f64().unwrap_or(0.0),
        cls: parsed["cls"].as_f64().unwrap_or(0.0),
        last_task_end: parsed["lastTaskEnd"].as_f64().unwrap_or(0.0),
        cls_shifts,
        inp_duration: parsed["inpDuration"].as_f64().unwrap_or(0.0),
    })
}

/// Parse CLS shift entries from the JSON value returned by the browser.
fn parse_cls_shifts(val: &serde_json::Value) -> Vec<ClsShift> {
    val.as_array()
        .map(|arr| {
            arr.iter()
                .filter_map(|entry| {
                    let value = entry["value"].as_f64()?;
                    let start_time_ms = entry["startTime"].as_f64().unwrap_or(0.0);
                    let sources = entry["sources"]
                        .as_array()
                        .map(|srcs| {
                            srcs.iter()
                                .map(|s| {
                                    let node = s["node"].as_str().unwrap_or("unknown").to_string();
                                    let previous_rect = parse_shift_rect(&s["previousRect"]);
                                    let current_rect = parse_shift_rect(&s["currentRect"]);
                                    ClsSource {
                                        node,
                                        previous_rect,
                                        current_rect,
                                    }
                                })
                                .collect()
                        })
                        .unwrap_or_default();
                    Some(ClsShift {
                        value,
                        start_time_ms,
                        sources,
                    })
                })
                .collect()
        })
        .unwrap_or_default()
}

/// Cumulative Layout Shift per the current Core Web Vitals definition: the
/// **maximum session window**, not the cumulative sum of every shift.
///
/// A session window groups consecutive layout shifts that are at most 1 s apart
/// and span at most 5 s in total; CLS is the largest window's summed value.
/// Summing every shift over the whole observation window (the pre-2020
/// definition) badly overcounts long-lived pages that shift repeatedly during
/// lazy-loading — which is exactly what a headless audit observes. This matches
/// Chrome's `web-vitals` library and Lighthouse.
pub(super) fn session_window_cls(shifts: &[ClsShift]) -> f64 {
    let mut ordered: Vec<&ClsShift> = shifts.iter().collect();
    ordered.sort_by(|a, b| a.start_time_ms.total_cmp(&b.start_time_ms));

    let mut max_window = 0.0_f64;
    let mut window_sum = 0.0_f64;
    let mut window_start = 0.0_f64;
    let mut prev_ts = 0.0_f64;
    let mut in_window = false;

    for shift in ordered {
        let ts = shift.start_time_ms;
        if !in_window || ts - prev_ts > 1000.0 || ts - window_start > 5000.0 {
            window_sum = shift.value;
            window_start = ts;
            in_window = true;
        } else {
            window_sum += shift.value;
        }
        prev_ts = ts;
        if window_sum > max_window {
            max_window = window_sum;
        }
    }
    max_window
}

/// Re-tag a vitals set's directly-measured metrics (LCP, FCP, CLS, TBT) as
/// captured under the Lighthouse mobile throttle. Estimated metrics (TTI,
/// Speed Index, INP) keep their `estimated_lab` provenance. Applied when the
/// throttled LhMobile pass is adopted as the canonical report measurement, so
/// the JSON honestly reflects that the headline vitals are throttled.
pub fn mark_throttled_mobile(vitals: &mut WebVitals) {
    fn tag(slot: &mut Option<VitalMetric>) {
        if let Some(metric) = slot.take() {
            *slot = Some(metric.into_throttled_mobile());
        }
    }
    tag(&mut vitals.lcp);
    tag(&mut vitals.fcp);
    tag(&mut vitals.cls);
    tag(&mut vitals.tbt);
}

fn parse_shift_rect(val: &serde_json::Value) -> Option<ShiftRect> {
    if val.is_null() {
        return None;
    }
    Some(ShiftRect {
        top: val["top"].as_f64().unwrap_or(0.0),
        left: val["left"].as_f64().unwrap_or(0.0),
        bottom: val["bottom"].as_f64().unwrap_or(0.0),
        right: val["right"].as_f64().unwrap_or(0.0),
    })
}

/// Values collected by the pre-injected PerformanceObserver script.
struct PreinjectedVitals {
    lcp: f64,
    tbt: f64,
    cls: f64,
    /// End time of the last long task (ms from navigation start); 0 = no long tasks
    last_task_end: f64,
    /// Per-shift CLS entries (#135)
    cls_shifts: Vec<ClsShift>,
    /// Max event/interaction duration captured by event-timing or first-input (#134)
    inp_duration: f64,
}

/// Read the current LCP value from the pre-injected global.
///
/// Returns `None` if no LCP was captured yet or the script was not injected.
pub async fn finalize_lcp(page: &Page) -> Result<Option<f64>> {
    let result = page
        .evaluate("window.__ams_lcp || 0")
        .await
        .map_err(|e| AuditError::CdpError(format!("LCP read failed: {}", e)))?;

    let val = result.value().and_then(|v| v.as_f64()).unwrap_or(0.0);
    Ok(if val > 0.0 { Some(val) } else { None })
}

/// Wait for LCP and TBT to stabilize before reading vitals.
///
/// **Phase 1 – LCP**: polls `window.__ams_lcp` every 250 ms (up to ~3 s)
/// until an LCP entry is captured.
///
/// **Phase 2 – TBT settle**: JS bundles execute *after* LCP fires, so Long Tasks
/// are not yet visible when LCP > 0. Polls `window.__ams_tbt` every 200 ms and
/// breaks only when the value is stable for three consecutive reads (≥ 600 ms
/// unchanged). Max 5 s — matches Lighthouse's quiet-window heuristic.
async fn wait_for_vitals_stable(page: &Page) {
    // Phase 1: wait for first LCP entry
    for _ in 0..11u8 {
        let lcp = page
            .evaluate("window.__ams_lcp || 0")
            .await
            .ok()
            .and_then(|r| r.value().and_then(|v| v.as_f64()))
            .unwrap_or(0.0);
        if lcp > 0.0 {
            break;
        }
        tokio::time::sleep(std::time::Duration::from_millis(250)).await;
    }

    // Phase 2: wait for TBT to stop growing (Long Tasks settle after LCP)
    let mut last_tbt = -1.0_f64;
    let mut stable_ticks: u8 = 0;
    for _ in 0..25u8 {
        let tbt = page
            .evaluate("window.__ams_tbt || 0")
            .await
            .ok()
            .and_then(|r| r.value().and_then(|v| v.as_f64()))
            .unwrap_or(0.0);
        if (tbt - last_tbt).abs() < 1.0 {
            stable_ticks += 1;
            if stable_ticks >= 3 {
                debug!(
                    "TBT stable at {:.0}ms after {} settle ticks",
                    tbt, stable_ticks
                );
                break;
            }
        } else {
            stable_ticks = 0;
        }
        last_tbt = tbt;
        tokio::time::sleep(std::time::Duration::from_millis(200)).await;
    }
}

/// Extract Core Web Vitals from a page
///
/// # Arguments
/// * `page` - The chromiumoxide Page to extract metrics from
///
/// # Returns
/// * `Ok(WebVitals)` - The extracted performance metrics
/// * `Err(AuditError)` - If extraction fails
pub async fn extract_web_vitals(page: &Page) -> Result<WebVitals> {
    info!("Extracting Core Web Vitals...");

    wait_for_vitals_stable(page).await;

    let mut vitals = WebVitals::default();

    // Read pre-injected observer globals first (Lighthouse-style, captures from first paint).
    // Falls back gracefully when prepare_vitals_collection was not called.
    let preinjected = collect_preinjected_vitals(page).await.unwrap_or_else(|e| {
        warn!(
            "Pre-injected vitals unavailable ({}), falling back to CDP",
            e
        );
        PreinjectedVitals {
            lcp: 0.0,
            tbt: 0.0,
            cls: 0.0,
            last_task_end: 0.0,
            cls_shifts: vec![],
            inp_duration: 0.0,
        }
    });
    let last_task_end = preinjected.last_task_end;

    // Apply pre-injected LCP and TBT when non-zero (0 = not captured / not injected)
    if preinjected.lcp > 0.0 && preinjected.lcp < 300_000.0 {
        vitals.lcp = Some(VitalMetric::new(preinjected.lcp, 2500.0, 4000.0));
        debug!("LCP (preinjected): {:.0}ms", preinjected.lcp);
    }
    // TBT: 0 ms is a valid perfect score (no long tasks), always include like CLS.
    vitals.tbt = Some(VitalMetric::new(preinjected.tbt, 200.0, 600.0));
    debug!("TBT (preinjected): {:.0}ms", preinjected.tbt);
    // CLS: 0.0 is a valid perfect score, always apply. Use the session-window
    // maximum (the current CWV definition) computed from the per-shift list,
    // not the cumulative sum the observer accumulates — the latter overcounts
    // pages that shift repeatedly during lazy-loading.
    let cls_session = session_window_cls(&preinjected.cls_shifts);
    vitals.cls = Some(VitalMetric::new(cls_session, 0.1, 0.25));
    debug!(
        "CLS session-window: {:.4} (cumulative was {:.4})",
        cls_session, preinjected.cls
    );

    // CLS attribution (#135)
    vitals.cls_attribution = preinjected.cls_shifts;

    // INP (#134): event-timing or first-input duration; 0 = no interaction captured.
    // In headless audits interactions rarely fire, so 0 means "not measurable" and
    // we leave the field empty rather than reporting a misleading 0ms.
    if preinjected.inp_duration > 0.0 {
        vitals.inp = Some(VitalMetric::new(preinjected.inp_duration, 200.0, 500.0).estimated());
        debug!("INP (event-timing): {:.0}ms", preinjected.inp_duration);
    }

    // Get performance metrics via CDP for FCP, DCL, heap, node count
    let metrics_response = page
        .execute(GetMetricsParams::default())
        .await
        .map_err(|e| AuditError::CdpError(format!("Failed to get metrics: {}", e)))?;

    for metric in &metrics_response.metrics {
        let value = metric.value;
        match metric.name.as_str() {
            "FirstContentfulPaint" => {
                let ms = value * 1000.0;
                if ms > 0.0 && ms < 300_000.0 {
                    vitals.fcp = Some(VitalMetric::new(ms, 1800.0, 3000.0));
                    debug!("FCP (CDP): {:.0}ms", ms);
                }
            }
            "LargestContentfulPaint" if vitals.lcp.is_none() => {
                // Only use CDP LCP as fallback when pre-injected observer missed it
                let ms = value * 1000.0;
                if ms > 0.0 && ms < 300_000.0 {
                    vitals.lcp = Some(VitalMetric::new(ms, 2500.0, 4000.0));
                    debug!("LCP (CDP fallback): {:.0}ms", ms);
                }
            }
            "DomContentLoaded" => {
                let ms = value * 1000.0;
                if ms > 0.0 && ms < 300_000.0 {
                    vitals.dom_content_loaded = Some(ms);
                }
            }
            "JSHeapUsedSize" => {
                vitals.js_heap_size = Some(value as i64);
            }
            "Nodes" => {
                vitals.dom_nodes = Some(value as i64);
            }
            _ => {}
        }
    }

    // Extract additional metrics via JavaScript (FCP fallback, TTFB, load time)
    let js_metrics = extract_js_metrics(page).await?;

    if vitals.fcp.is_none() && js_metrics.fcp.is_some() {
        vitals.fcp = js_metrics.fcp;
    }
    if vitals.lcp.is_none() && js_metrics.lcp.is_some() {
        vitals.lcp = js_metrics.lcp;
    }
    if js_metrics.ttfb.is_some() {
        vitals.ttfb = js_metrics.ttfb;
    }
    if js_metrics.load_time.is_some() {
        vitals.load_time = js_metrics.load_time;
    }
    if vitals.dom_content_loaded.is_none() {
        if let Some(js_dcl) = js_metrics.dom_content_loaded {
            if js_dcl > 0.0 && js_dcl < 300_000.0 {
                vitals.dom_content_loaded = Some(js_dcl);
            }
        }
    }

    // TTI (#137): max(domInteractive, last_long_task_end).
    // domInteractive = domContentLoadedEventStart (navigation timing).
    // Thresholds from Lighthouse: good ≤3800ms, poor >7300ms.
    let dom_interactive_ms: f64 = page
        .evaluate("(function(){ var n = performance.getEntriesByType('navigation')[0]; return n ? n.domInteractive : 0; })()")
        .await
        .ok()
        .and_then(|r| r.value().and_then(|v| v.as_f64()))
        .unwrap_or(0.0);
    let mut tti_ms = dom_interactive_ms.max(last_task_end);
    if let Some(ref lcp_metric) = vitals.lcp {
        tti_ms = tti_ms.max(lcp_metric.value);
    }
    if tti_ms > 0.0 && tti_ms < 300_000.0 {
        vitals.tti = Some(VitalMetric::new(tti_ms, 3800.0, 7300.0).estimated());
        debug!("TTI (approx): {:.0}ms", tti_ms);
    }

    // Speed Index (#137): heuristic proxy = 0.35·FCP + 0.65·LCP.
    // Real Speed Index requires video-frame analysis; this approximation
    // correlates well for typical content-heavy pages.
    // Thresholds from Lighthouse: good ≤3400ms, poor >5800ms.
    if let (Some(fcp), Some(lcp)) = (&vitals.fcp, &vitals.lcp) {
        let si = 0.35 * fcp.value + 0.65 * lcp.value;
        vitals.speed_index = Some(VitalMetric::new(si, 3400.0, 5800.0).estimated());
        debug!("Speed Index (heuristic): {:.0}ms", si);
    }

    info!(
        "Web Vitals extracted: LCP={:?}ms, FCP={:?}ms, CLS={:?}, TTI={:?}ms, INP={:?}ms",
        vitals.lcp.as_ref().map(|v| v.value as i64),
        vitals.fcp.as_ref().map(|v| v.value as i64),
        vitals.cls.as_ref().map(|v| v.value),
        vitals.tti.as_ref().map(|v| v.value as i64),
        vitals.inp.as_ref().map(|v| v.value as i64),
    );

    Ok(vitals)
}

/// Extract metrics via JavaScript Performance API
async fn extract_js_metrics(page: &Page) -> Result<WebVitals> {
    let js_code = r#"
    (() => {
        const result = {};

        // Navigation Timing
        const nav = performance.getEntriesByType('navigation')[0];
        if (nav) {
            result.ttfb = nav.responseStart - nav.requestStart;
            if (nav.loadEventEnd > 0) {
                result.loadTime = nav.loadEventEnd - nav.startTime;
            }
            if (nav.domContentLoadedEventEnd > 0) {
                result.domContentLoaded = nav.domContentLoadedEventEnd - nav.startTime;
            }
        }

        // Paint Timing
        const paints = performance.getEntriesByType('paint');
        for (const paint of paints) {
            if (paint.name === 'first-contentful-paint') {
                result.fcp = paint.startTime;
            }
        }

        // LCP (if available)
        const lcpEntries = performance.getEntriesByType('largest-contentful-paint');
        if (lcpEntries.length > 0) {
            result.lcp = lcpEntries[lcpEntries.length - 1].startTime;
        }

        return JSON.stringify(result);
    })()
    "#;

    let js_result = page
        .evaluate(js_code)
        .await
        .map_err(|e| AuditError::CdpError(format!("JS metrics extraction failed: {}", e)))?;

    let json_str = js_result.value().and_then(|v| v.as_str()).unwrap_or("{}");

    let parsed: serde_json::Value = serde_json::from_str(json_str).unwrap_or_default();

    let mut vitals = WebVitals::default();

    if let Some(lcp) = parsed["lcp"].as_f64() {
        vitals.lcp = Some(VitalMetric::new(lcp, 2500.0, 4000.0));
    }
    if let Some(fcp) = parsed["fcp"].as_f64() {
        vitals.fcp = Some(VitalMetric::new(fcp, 1800.0, 3000.0));
    }
    if let Some(ttfb) = parsed["ttfb"].as_f64() {
        vitals.ttfb = Some(VitalMetric::new(ttfb, 800.0, 1800.0));
    }
    if let Some(load_time) = parsed["loadTime"].as_f64() {
        vitals.load_time = Some(load_time);
    }
    if let Some(dcl) = parsed["domContentLoaded"].as_f64() {
        vitals.dom_content_loaded = Some(dcl);
    }

    Ok(vitals)
}

/// Detect implausible or structurally unmeasurable metrics in a headless CDP audit.
///
/// Returns a list of warning keys that document known limitations so consumers
/// can adjust score interpretation or surface the caveats in reports.
///
/// Warning keys (issue #291):
/// - `"lcp_not_measured"` — LCP absent (observer missed it, e.g. under LhMobile throttle)
/// - `"tbt_zero_heavy_page"` — TBT=0 while LCP>3000ms (physically impossible)
/// - `"speed_index_fallback_to_lcp"` — Speed Index equals LCP (no independent value)
/// - `"tti_fallback_to_lcp"` — TTI equals LCP (no independent value)
/// - `"inp_not_measured"` — INP absent (headless has no user interaction)
pub fn validate_metrics(vitals: &WebVitals) -> Vec<String> {
    let mut warnings = Vec::new();

    if vitals.lcp.is_none() {
        warnings.push("lcp_not_measured".to_string());
    }

    if let (Some(tbt), Some(lcp)) = (&vitals.tbt, &vitals.lcp) {
        if tbt.value == 0.0 && lcp.value > 3000.0 {
            warnings.push("tbt_zero_heavy_page".to_string());
        }
    }

    if let (Some(si), Some(lcp)) = (&vitals.speed_index, &vitals.lcp) {
        if (si.value - lcp.value).abs() < f64::EPSILON {
            warnings.push("speed_index_fallback_to_lcp".to_string());
        }
    }

    if let (Some(tti), Some(lcp)) = (&vitals.tti, &vitals.lcp) {
        if (tti.value - lcp.value).abs() < f64::EPSILON {
            warnings.push("tti_fallback_to_lcp".to_string());
        }
    }

    if vitals.inp.is_none() {
        warnings.push("inp_not_measured".to_string());
    }

    warnings
}

#[cfg(test)]
mod tests {
    use super::*;

    #[test]
    fn test_vital_metric_good() {
        let metric = VitalMetric::new(1500.0, 2500.0, 4000.0);
        assert!(metric.is_good());
        assert_eq!(metric.rating, "good");
    }

    #[test]
    fn test_vital_metric_needs_improvement() {
        let metric = VitalMetric::new(3000.0, 2500.0, 4000.0);
        assert!(!metric.is_good());
        assert_eq!(metric.rating, "needs-improvement");
    }

    #[test]
    fn test_vital_metric_poor() {
        let metric = VitalMetric::new(5000.0, 2500.0, 4000.0);
        assert!(!metric.is_good());
        assert_eq!(metric.rating, "poor");
    }

    #[test]
    fn test_measurement_context_defaults_to_lab_headless() {
        let metric = VitalMetric::new(1500.0, 2500.0, 4000.0);
        assert_eq!(metric.measurement, MeasurementContext::LabHeadless);
        assert!(!metric.is_estimated());

        let json = serde_json::to_string(&metric).unwrap();
        assert!(json.contains("\"measurement\":\"lab_headless\""));
    }

    #[test]
    fn test_estimated_marks_metric_as_estimated_lab() {
        let metric = VitalMetric::new(3000.0, 3800.0, 7300.0).estimated();
        assert_eq!(metric.measurement, MeasurementContext::EstimatedLab);
        assert!(metric.is_estimated());

        let json = serde_json::to_string(&metric).unwrap();
        assert!(json.contains("\"measurement\":\"estimated_lab\""));
    }

    #[test]
    fn test_web_vitals_good_count() {
        let vitals = WebVitals {
            lcp: Some(VitalMetric::new(2000.0, 2500.0, 4000.0)),
            fcp: Some(VitalMetric::new(1500.0, 1800.0, 3000.0)),
            cls: Some(VitalMetric::new(0.05, 0.1, 0.25)),
            ..Default::default()
        };

        assert_eq!(vitals.good_count(), 3);
        assert_eq!(vitals.available_count(), 3);
    }

    fn shift(value: f64, start_time_ms: f64) -> ClsShift {
        ClsShift {
            value,
            start_time_ms,
            sources: vec![],
        }
    }

    #[test]
    fn session_window_cls_matches_cwv_definition() {
        // No shifts -> 0.
        assert_eq!(session_window_cls(&[]), 0.0);

        // Shifts within one 5 s window, <1 s apart -> summed.
        let one_window = [shift(0.1, 500.0), shift(0.2, 1000.0), shift(0.05, 1500.0)];
        assert!((session_window_cls(&one_window) - 0.35).abs() < 1e-9);

        // A >1 s gap splits windows; CLS is the MAX window, not the sum.
        // Window A = 0.1+0.1 = 0.2; gap; Window B = 0.4 -> max 0.4 (cumulative
        // sum would be 0.6).
        let two_windows = [
            shift(0.1, 0.0),
            shift(0.1, 500.0),
            shift(0.4, 3000.0), // 2.5 s after previous shift -> new window
        ];
        assert!((session_window_cls(&two_windows) - 0.4).abs() < 1e-9);

        // Shifts spaced <1 s apart but spanning >5 s start a fresh window once
        // the 5 s cap is exceeded.
        let long_run: Vec<ClsShift> = (0..10).map(|i| shift(0.1, i as f64 * 800.0)).collect();
        // Window 1 covers 0..5000ms = 7 shifts (0,800,..,4800) = 0.7; the cap
        // resets at 5600ms, so a single window never sums all ten.
        let cls = session_window_cls(&long_run);
        assert!(cls < 1.0, "expected windowed value <1.0, got {cls}");
        assert!(cls >= 0.7 - 1e-9, "first window should sum ~0.7, got {cls}");
    }

    #[test]
    fn mark_throttled_mobile_tags_direct_metrics_only() {
        let mut v = WebVitals {
            lcp: Some(VitalMetric::new(3800.0, 2500.0, 4000.0)),
            cls: Some(VitalMetric::new(0.05, 0.1, 0.25)),
            tbt: Some(VitalMetric::new(450.0, 200.0, 600.0)),
            // Estimated metric: must keep its estimated provenance.
            speed_index: Some(VitalMetric::new(4200.0, 3400.0, 5800.0).estimated()),
            ..Default::default()
        };
        mark_throttled_mobile(&mut v);
        assert_eq!(
            v.lcp.unwrap().measurement,
            MeasurementContext::LabThrottledMobile
        );
        assert_eq!(
            v.cls.unwrap().measurement,
            MeasurementContext::LabThrottledMobile
        );
        assert_eq!(
            v.tbt.unwrap().measurement,
            MeasurementContext::LabThrottledMobile
        );
        // Estimated stays estimated.
        assert_eq!(
            v.speed_index.unwrap().measurement,
            MeasurementContext::EstimatedLab
        );
    }

    #[test]
    fn validate_metrics_tbt_zero_heavy_page() {
        let vitals = WebVitals {
            tbt: Some(VitalMetric::new(0.0, 200.0, 600.0)),
            lcp: Some(VitalMetric::new(4500.0, 2500.0, 4000.0)),
            ..Default::default()
        };
        let warnings = validate_metrics(&vitals);
        assert!(
            warnings.contains(&"tbt_zero_heavy_page".to_string()),
            "must warn when TBT=0 and LCP>3000ms"
        );
        assert!(
            warnings.contains(&"inp_not_measured".to_string()),
            "inp_not_measured always present in headless"
        );
    }

    #[test]
    fn validate_metrics_speed_index_fallback() {
        let lcp_val = 3200.0;
        let vitals = WebVitals {
            lcp: Some(VitalMetric::new(lcp_val, 2500.0, 4000.0)),
            speed_index: Some(VitalMetric::new(lcp_val, 3800.0, 5800.0).estimated()),
            ..Default::default()
        };
        let warnings = validate_metrics(&vitals);
        assert!(
            warnings.contains(&"speed_index_fallback_to_lcp".to_string()),
            "must warn when SpeedIndex == LCP"
        );
    }

    #[test]
    fn validate_metrics_no_warnings_on_good_data() {
        let vitals = WebVitals {
            tbt: Some(VitalMetric::new(120.0, 200.0, 600.0)),
            lcp: Some(VitalMetric::new(2000.0, 2500.0, 4000.0)),
            speed_index: Some(VitalMetric::new(1800.0, 3800.0, 5800.0).estimated()),
            tti: Some(VitalMetric::new(3500.0, 3800.0, 7300.0).estimated()),
            inp: Some(VitalMetric::new(150.0, 200.0, 500.0).estimated()),
            ..Default::default()
        };
        let warnings = validate_metrics(&vitals);
        assert!(
            !warnings.contains(&"tbt_zero_heavy_page".to_string()),
            "no tbt_zero_heavy_page when TBT>0"
        );
        assert!(
            !warnings.contains(&"speed_index_fallback_to_lcp".to_string()),
            "no speed_index_fallback when SI != LCP"
        );
        assert!(
            !warnings.contains(&"inp_not_measured".to_string()),
            "no inp_not_measured when INP is present"
        );
    }

    #[test]
    fn validate_metrics_inp_always_warned_in_headless() {
        let vitals = WebVitals::default(); // no INP
        let warnings = validate_metrics(&vitals);
        assert!(
            warnings.contains(&"inp_not_measured".to_string()),
            "inp_not_measured must be present when inp is None"
        );
    }

    #[test]
    fn validate_metrics_lcp_not_measured_when_absent() {
        let vitals = WebVitals::default(); // no LCP
        let warnings = validate_metrics(&vitals);
        assert!(
            warnings.contains(&"lcp_not_measured".to_string()),
            "lcp_not_measured must be present when lcp is None (#285)"
        );
    }

    #[test]
    fn validate_metrics_no_lcp_warning_when_lcp_present() {
        let vitals = WebVitals {
            lcp: Some(VitalMetric::new(2000.0, 2500.0, 4000.0)),
            ..Default::default()
        };
        let warnings = validate_metrics(&vitals);
        assert!(
            !warnings.contains(&"lcp_not_measured".to_string()),
            "no lcp_not_measured when lcp is present"
        );
    }
}