"""Rendu HTML « Recherchabilité fuzzy » — Sprint 86 (A.II.5a HTML).
Suite directe ``picarones/core/searchability.py`` (Sprint 84) +
câblage runner (Sprint 86).
Pattern identique aux autres rendus (Sprints 41/43/62/67/72) :
**server-side**, pas de JavaScript, anti-injection systématique.
Vue
---
Tableau résumé : moteur × (rappel, n_searchable / n_gt_tokens,
docs). Cellule rappel colorée par gradient rouge → vert.
Adaptative : ``""`` si aucun moteur n'a de
``aggregated_searchability``.
"""
from __future__ import annotations
from html import escape as _e
from typing import Optional
def _color_for_recall(recall: float) -> str:
"""Gradient rouge → jaune → vert pour rappel ∈ [0, 1]."""
f = max(0.0, min(1.0, recall))
if f < 0.5:
# rouge → jaune
t = f / 0.5
r = 235
g = int(70 + (200 - 70) * t)
b = 70
else:
# jaune → vert
t = (f - 0.5) / 0.5
r = int(235 + (60 - 235) * t)
g = int(200 + (160 - 200) * t)
b = int(70 + (90 - 70) * t)
return f"#{r:02x}{g:02x}{b:02x}"
def build_searchability_summary_html(
engines: list[dict],
labels: Optional[dict[str, str]] = None,
) -> str:
"""Construit la table HTML de recherchabilité.
Parameters
----------
engines:
Liste de dicts moteur ; chacun peut avoir
``aggregated_searchability``.
labels:
Dict i18n, clés ``search_*``.
Returns
-------
str
``""`` si aucun moteur n'a de signal.
"""
rows = [
e for e in engines
if isinstance(e.get("aggregated_searchability"), dict)
]
if not rows:
return ""
labels = labels or {}
title = labels.get("search_title", "Recherchabilité fuzzy")
note = labels.get(
"search_note",
"Proportion de tokens GT retrouvés dans la sortie OCR à "
"distance de Levenshtein ≤ 2 — proxy direct de la "
"qualité pour la recherche plein-texte (Elastic, Solr).",
)
col_engine = labels.get("search_engine", "Moteur")
col_recall = labels.get("search_recall", "Rappel")
col_count = labels.get("search_count", "Tokens retrouvés / total")
col_docs = labels.get("search_docs", "Docs")
parts = [
'',
f'
{_e(title)}
',
f'
'
f'{_e(note)}
',
'
',
'',
]
for col in (col_engine, col_recall, col_count, col_docs):
parts.append(
f'| '
f'{_e(col)} | '
)
parts.append("
")
for engine in rows:
agg = engine["aggregated_searchability"]
name = engine.get("name") or "?"
recall = float(agg.get("recall") or 0.0)
n_search = int(agg.get("n_searchable") or 0)
n_total = int(agg.get("n_gt_tokens") or 0)
n_docs = int(agg.get("n_docs") or 0)
color = _color_for_recall(recall)
parts.append(
f''
f'| {_e(str(name))} | '
f''
f'{recall * 100:.1f}% | '
f'{n_search} / {n_total} | '
f'{n_docs} | '
f'
'
)
parts.append("