{% extends "base.html" %}
{% block title_suffix %} — Baseline Status{% endblock %}
{% block content %}

<h1>Baseline Status</h1>

{# Honesty element 5: relative-not-absolute framing — prominent at top #}
<div class="caveat">
  Self-baseline verdicts are <strong>relative to YOUR OWN lean waste-free sessions of each type</strong>
  &mdash; not an absolute efficiency standard. "within_band" means comparable to your leaner sessions
  of that type, not that your work is universally efficient.
</div>

{% if headline %}
<div class="section">
  <div class="section-label">Coverage &mdash; Content Sessions</div>
  <div class="metric-box">
    Sessions with content:&nbsp;<strong>{{ headline.content_sessions }}</strong>
    &nbsp;&middot;&nbsp;
    Scored vs your lean runs:&nbsp;<span class="metric-highlight">{{ headline.content_self_scored }} ({{ headline.content_self_pct }}%)</span>
    &nbsp;&middot;&nbsp;
    Out of scope:&nbsp;{{ headline.content_sessions - headline.content_self_scored }}
    <br>
    <span class="metric-old">Pre-self-baseline (B2 corpus): {{ headline.b2_content_scored }} of {{ headline.b2_content_total }} ({{ headline.b2_content_scored_pct }}%)</span>
    <br>
    <em style="font-size:0.82em; color:#6b7280;">
      Empty/stub sessions (no work product): {{ headline.empty_stubs }} ({{ headline.empty_stubs_pct }}% of {{ headline.total }} total) &mdash; not scoring failures.
    </em>
  </div>
</div>
{% endif %}

{% if status_rows %}
<div class="section">
  <div class="section-label">Per-Type Baseline Status</div>
  <table>
    <thead>
      <tr>
        <th>Task Type</th>
        <th>Sessions</th>
        <th>In-Scope</th>
        <th>Lean-N</th>
        <th>Scope Floor</th>
        <th>Status</th>
        <th>Band</th>
      </tr>
    </thead>
    <tbody>
      {% for row in status_rows %}
      <tr>
        <td><strong>{{ row.task_type }}</strong></td>
        <td>{{ row.total }}</td>
        <td>{{ row.in_scope }}</td>
        <td>{{ row.lean_n }}</td>
        <td>{{ row.scope_floor }} turns</td>
        <td>
          {% if row.source == "self" %}
            <span class="badge badge-self">ACTIVE (self)</span>
          {% elif row.source == "building" %}
            <span class="badge badge-building">building</span>
            <span class="text-faint" style="font-size:0.78rem; margin-left:0.25rem;">{{ row.sessions_needed }} more needed</span>
          {% else %}
            <span class="badge badge-corpus">corpus fallback</span>
          {% endif %}
        </td>
        <td class="text-muted" style="font-size:0.82rem;">
          {% if row.p25 is not none and row.p75 is not none %}
            {{ "{:,}".format(row.p25) }} &ndash; {{ "{:,}".format(row.p75) }}
            <span class="text-faint" style="font-size:0.78rem;">(med {{ "{:,}".format(row.median) }})</span>
          {% else %}
            <span class="text-faint">&mdash;</span>
          {% endif %}
        </td>
      </tr>
      {% endfor %}
    </tbody>
  </table>
</div>

<div class="section">
  <div class="section-label">Domain of Validity per Type</div>
  {% for row in status_rows %}
  <p style="font-size:0.82rem; font-weight:600; color:#374151; margin-top:0.75rem; margin-bottom:0.2rem;">{{ row.task_type }}</p>
  <div class="caveat">{{ row.domain_of_validity }}</div>
  {% endfor %}
</div>

{% if waste_by_type %}
<div class="section">
  <div class="section-label">Waste Concentration</div>
  {% set total_events = waste_by_type | sum(attribute='total_events') %}
  {% if total_events > 0 %}
  <div class="caveat">
    Detectable waste (observable-invariant only: retry loops + redundant reads) is concentrated in
    <strong>ml-eval, infra-deploy, and research-recon</strong> &mdash; the infrastructure/ML-ops types.
    {% for row in waste_by_type if row.total_events > 0 %}
      <strong>{{ row.task_type }}</strong>: {{ row.total_events }} event(s) in {{ row.with_waste }} of {{ row.total }} sessions;
    {% endfor %}
    feature-build ({{ waste_by_type | selectattr('task_type', 'eq', 'feature-build') | map(attribute='total') | first | default(0) }} sessions): zero waste events.
    Total: {{ total_events }} events.
    See <a href="https://github.com/gaurav-gandhi-2411/token-efficiency-scorer/blob/master/research/11-generalization.md">report 11</a> for corpus characterisation.
  </div>
  {% else %}
  <div class="caveat">
    No waste events in store &mdash; run <code>tes backfill-waste</code> to detect and record waste
    across all sessions. Expected concentration: ml-eval &gt; infra-deploy &gt; research-recon; zero in feature-build.
  </div>
  {% endif %}
</div>
{% endif %}

{% else %}
<p class="empty">No baseline data available. Run <code>tes score</code> to accumulate sessions.</p>
{% endif %}

{% endblock %}