Synthèse factuelle
- - Générée mécaniquement depuis les résultats — aucun LLM, reproductible. - --
- {% for sentence in synthesis.sentences %}
-
- {{ sentence }} - {% endfor %} -
- Pour comprendre comment d'autres équipes ont raisonné sur des problèmes - similaires, voir - les études de cas. -
-
-
diff --git a/picarones/report/templates/view_characters.html b/picarones/report/templates/view_characters.html
deleted file mode 100644
index 9641eeacf35fec0a17eee05c946f58b309b78e9e..0000000000000000000000000000000000000000
--- a/picarones/report/templates/view_characters.html
+++ /dev/null
@@ -1,32 +0,0 @@
-
-
-
-
-
-
-
- Distribution du CER par moteur
-
-
-
-
-
-
- Profil des moteurs (radar)
-
-
-
-
- Axe radar : CER, WER, MER, WIL — valeurs inversées (plus c'est haut, meilleur est le moteur).
-
-
-
-
- CER par document (tous moteurs)
-
-
-
-
-
-
- Temps d'exécution moyen (secondes/document)
-
-
-
-
-
-
- Qualité image ↔ CER (scatter plot)
-
-
-
-
- Chaque point = un document. Axe X = score qualité image [0–1]. Axe Y = CER. Corrélation négative attendue.
-
-
-
-
-
- Taxonomie des erreurs par moteur
-
-
-
-
- Distribution des classes d'erreurs (classes 1–9 de la taxonomie Picarones).
-
-
-
-
-
- Courbes de fiabilité
-
-
-
-
- Pour les X% documents les plus faciles (triés par CER croissant), quel est le CER moyen cumulé ?
- Une courbe basse = moteur performant même sur les documents faciles.
-
-
-
-
-
- Intervalles de confiance à 95 % (bootstrap)
-
-
-
-
- IC à 95% sur le CER moyen par moteur (1000 itérations bootstrap).
-
-
-
-
-
- Erreurs communes / exclusives (Venn)
- -
- Intersection des ensembles d'erreurs entre les 2 ou 3 premiers concurrents.
- Erreurs communes = segments partagés.
-
-
-
-
-
- Tests de Wilcoxon — comparaisons par paires
- -
- Test signé-rangé de Wilcoxon (non-paramétrique). Seuil α = 0.05.
-
-
-
-
-
- Clustering des patterns d'erreurs
- -
-
-
-
- Gini vs CER moyen — idéal : bas-gauche
-
-
-
-
- Axe X = CER moyen, Axe Y = coefficient de Gini. Un moteur idéal a CER bas ET Gini bas (erreurs rares et uniformes).
-
-
-
-
-
- Ratio longueur vs ancrage — hallucinations VLM
-
-
-
-
- Axe X = score d'ancrage trigrammes [0–1]. Axe Y = ratio longueur sortie/GT.
- Zone ⚠️ : ancrage < 0.5 ou ratio > 1.2 → hallucinations probables.
-
-
-
-
-
- {% if calibration_summary_html or reliability_diagrams_html %}
- Compromis qualité / coût
- - -
- Les moteurs sur la frontière de Pareto (en évidence) sont ceux pour
- lesquels aucun autre moteur n'offre simultanément un meilleur CER ET
- un meilleur coût. Prix indicatifs (table interne, datée). Le mode
- carbone est expérimental.
-
-
-
- Hypothèses détaillées par moteur
-
-
- {% endif %}
-
-
- {% if ner_summary_html or ner_per_category_html %}
- Calibration des moteurs
-
- {% if calibration_summary_html %}
-
- {{ calibration_summary_html }}
- {% endif %}
- {% if reliability_diagrams_html %}
- {{ reliability_diagrams_html }}
- {% endif %}
-
- ECE (Expected Calibration Error) : moyenne pondérée des écarts
- |confiance − précision| par bin. Plus l'ECE est bas, plus le
- moteur est honnête sur sa fiabilité — la diagonale du diagramme
- représente la calibration parfaite. Un ECE élevé signale qu'on
- ne peut pas se fier au score de confiance pour cibler la
- relecture humaine.
-
-
-
- {% endif %}
-
-
- {% if philological_profile_html %}
- Précision sur entités nommées
-
- {% if ner_summary_html %}
-
- {{ ner_summary_html }}
- {% endif %}
- {% if ner_per_category_html %}
- {{ ner_per_category_html }}
- {% endif %}
-
- F1 calculé par alignement IoU ≥ 0,5 sur les spans (labels
- case-insensitive). Plus le F1 est haut, plus le moteur restitue
- fidèlement les entités nommées (personnes, lieux, dates) — ce
- qui prédit l'utilité aval pour l'indexation prosopographique.
- Cette métrique mesure conjointement OCR + extracteur NER ; le
- modèle d'extraction lui-même peut halluciner.
-
-
- {{ philological_profile_html }}
-
- {% endif %}
-
-
- {% if searchability_html %}
-
- {{ searchability_html }}
-
- {% endif %}
- {% if numerical_sequences_html %}
-
- {{ numerical_sequences_html }}
-
- {% endif %}
-
-
- {% if readability_html %}
-
- {{ readability_html }}
-
- {% endif %}
-
-
- {% if specialization_html %}
-
- {{ specialization_html }}
-
- {% endif %}
-
-
- {% if divergence_matrix_html or oracle_gap_html %}
-
-
- {% endif %}
-
-
- {% if economics_view_html %}
- Analyse inter-moteurs
-
- {% if divergence_matrix_html %}
-
- {{ divergence_matrix_html }}
- {% endif %}
- {% if oracle_gap_html %}
- {{ oracle_gap_html }}
- {% endif %}
-
- Plus la divergence est élevée, plus deux moteurs se trompent sur des
- classes d'erreurs différentes — ils sont alors candidats à un voting
- ensemble. L'oracle est la borne supérieure du recall token-level
- atteignable par ce voting (proxy bag-of-words).
-
-
- {{ economics_view_html }}
-
- {% endif %}
- {% if advanced_taxonomy_view_html %}
-
- {{ advanced_taxonomy_view_html }}
-
- {% endif %}
- {% if diagnostics_view_html %}
-
- {{ diagnostics_view_html }}
-
- {% endif %}
-
-
- {% if rare_token_recall_html %}
-
- {{ rare_token_recall_html }}
-
- {% endif %}
- {% if taxonomy_cooccurrence_html %}
-
- {{ taxonomy_cooccurrence_html }}
-
- {% endif %}
- {% if taxonomy_intra_doc_html %}
-
- {{ taxonomy_intra_doc_html }}
-
- {% endif %}
- {% if marginal_cost_html %}
-
- {{ marginal_cost_html }}
-
- {% endif %}
-
-
-
-
-
- Matrice de corrélation entre métriques
-
-
-
-
-
- Coefficient de Pearson entre les métriques CER, WER, qualité image, ligatures, diacritiques.
- Vert = corrélation positive, Rouge = corrélation négative.
-
-
-
diff --git a/picarones/report/templates/view_document.html b/picarones/report/templates/view_document.html
deleted file mode 100644
index 65b653b63493733fc687c89c25f4a253c30c8dff..0000000000000000000000000000000000000000
--- a/picarones/report/templates/view_document.html
+++ /dev/null
@@ -1,83 +0,0 @@
-
-
-
-
-Analyse des caractères
- - -
-
-
-
-
-
-
-
-
- Matrice de confusion unicode - - — substitutions les plus fréquentes (caractère GT → caractère OCR) - -
- - - -Reconnaissance des ligatures
- - - -Distribution taxonomique des erreurs
- -
-
diff --git a/picarones/report/templates/view_gallery.html b/picarones/report/templates/view_gallery.html
deleted file mode 100644
index cbfec5e59d58a406753aae7a97711851b1241643..0000000000000000000000000000000000000000
--- a/picarones/report/templates/view_gallery.html
+++ /dev/null
@@ -1,35 +0,0 @@
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
- Sélectionner un document
- -
-
-
-
- Image originale
-
-
-
- 🖼
- Sélectionnez un document
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
- Comparaison GT / OCR
- -
-
-
-
-
-
-
- ✓ Vérité terrain (GT)
-
- —
-
-
-
- OCR
-
-
- —
-
-
diff --git a/picarones/report/templates/view_ranking.html b/picarones/report/templates/view_ranking.html
deleted file mode 100644
index 01dfdfa2ae509a9ecfac8aa321fd4590621b3490..0000000000000000000000000000000000000000
--- a/picarones/report/templates/view_ranking.html
+++ /dev/null
@@ -1,91 +0,0 @@
-
-
-
-
-Galerie des documents
-
-
-
-
-
-
-
-
-
-
-
diff --git a/picarones/reports_v2/html/__init__.py b/picarones/reports_v2/html/__init__.py
index 32cf1d15f16c9d9a418a003fee5fe550b0f43dd1..4bfeaa45ccb44c8b33e0c786f15828b7393287bb 100644
--- a/picarones/reports_v2/html/__init__.py
+++ b/picarones/reports_v2/html/__init__.py
@@ -21,6 +21,7 @@ Usage
from __future__ import annotations
+from picarones.reports_v2.html.generator import ReportGenerator
from picarones.reports_v2.html.render import HtmlReportRenderer
-__all__ = ["HtmlReportRenderer"]
+__all__ = ["HtmlReportRenderer", "ReportGenerator"]
diff --git a/picarones/reports_v2/html/comparison.py b/picarones/reports_v2/html/comparison.py
new file mode 100644
index 0000000000000000000000000000000000000000..ae03a544a3730a8df3bbe9b1323ff9a14ee02fb2
--- /dev/null
+++ b/picarones/reports_v2/html/comparison.py
@@ -0,0 +1,414 @@
+"""Comparaison de deux runs de benchmark (Sprint 28).
+
+Phase 5.E — module relocalisé depuis ``picarones.report.comparison``
+vers ``picarones.reports_v2.html.comparison``. Le chemin legacy
+reste disponible via un shim avec ``DeprecationWarning`` ;
+suppression prévue en 2.0.
+
+Le Sprint 8 a livré la persistance longitudinale via SQLite
+(``picarones.measurements.history``) et un détecteur de régression CLI. Mais
+aucun outil n'exposait la **comparaison** de deux runs côté rapport :
+un chercheur qui itère sur 8 prompts ne pouvait pas voir d'un coup
+*« Tesseract → GPT-4o version V2 a régressé de 0,8 pp en CER moyen
+sur la strate paroissiaux par rapport à V1 »*.
+
+Ce module fournit :
+
+- ``load_benchmark_json(path)`` — charge le JSON produit par
+ ``BenchmarkResult.as_dict()`` ou ``picarones run -o results.json``.
+- ``compare_benchmarks(a, b)`` — calcule les deltas par moteur
+ (CER mean, WER mean, comptes de documents traités/échoués) et
+ par strate quand la métadonnée est présente.
+- ``detect_regressions(diff, threshold)`` — liste les moteurs en
+ régression (delta CER > threshold) et en amélioration
+ (delta CER < -threshold).
+- ``render_comparison_html(diff, output_path)`` — rendu HTML
+ auto-contenu minimal via Jinja2 pour partage.
+
+Conventions
+-----------
+- Les deltas sont calculés ``b - a`` (donc positif = ``b`` est pire).
+- Un moteur présent dans un seul run apparaît dans ``only_in_a`` /
+ ``only_in_b``, jamais dans ``deltas``.
+- Un moteur dont le ``mean_cer`` est ``None`` (échec total) est
+ signalé mais ne génère pas de delta numérique.
+- ``threshold`` est en absolu (CER en fraction, pas en %). Défaut
+ 0.005 = 0,5 pp.
+"""
+
+from __future__ import annotations
+
+import json
+import logging
+from dataclasses import dataclass, field
+from pathlib import Path
+from typing import Any, Optional
+
+logger = logging.getLogger(__name__)
+
+
+# ---------------------------------------------------------------------------
+# Modèles
+# ---------------------------------------------------------------------------
+
+@dataclass
+class EngineDelta:
+ """Différence ``b - a`` pour un moteur donné."""
+ engine: str
+ cer_a: Optional[float]
+ cer_b: Optional[float]
+ delta_cer: Optional[float]
+ wer_a: Optional[float]
+ wer_b: Optional[float]
+ delta_wer: Optional[float]
+ docs_a: int
+ docs_b: int
+ failed_a: int
+ failed_b: int
+ is_regression: bool = False
+ is_improvement: bool = False
+
+ def as_dict(self) -> dict[str, Any]:
+ return {
+ "engine": self.engine,
+ "cer_a": self.cer_a,
+ "cer_b": self.cer_b,
+ "delta_cer": self.delta_cer,
+ "wer_a": self.wer_a,
+ "wer_b": self.wer_b,
+ "delta_wer": self.delta_wer,
+ "docs_a": self.docs_a,
+ "docs_b": self.docs_b,
+ "failed_a": self.failed_a,
+ "failed_b": self.failed_b,
+ "is_regression": self.is_regression,
+ "is_improvement": self.is_improvement,
+ }
+
+
+@dataclass
+class ComparisonResult:
+ """Résultat d'une comparaison ``b - a`` entre deux runs."""
+ label_a: str
+ label_b: str
+ run_date_a: Optional[str]
+ run_date_b: Optional[str]
+ corpus_a: Optional[str]
+ corpus_b: Optional[str]
+ deltas: list[EngineDelta] = field(default_factory=list)
+ only_in_a: list[str] = field(default_factory=list)
+ only_in_b: list[str] = field(default_factory=list)
+ threshold: float = 0.005
+
+ def as_dict(self) -> dict[str, Any]:
+ return {
+ "label_a": self.label_a,
+ "label_b": self.label_b,
+ "run_date_a": self.run_date_a,
+ "run_date_b": self.run_date_b,
+ "corpus_a": self.corpus_a,
+ "corpus_b": self.corpus_b,
+ "threshold": self.threshold,
+ "deltas": [d.as_dict() for d in self.deltas],
+ "only_in_a": list(self.only_in_a),
+ "only_in_b": list(self.only_in_b),
+ "regressions": [d.as_dict() for d in self.deltas if d.is_regression],
+ "improvements": [d.as_dict() for d in self.deltas if d.is_improvement],
+ }
+
+
+# ---------------------------------------------------------------------------
+# Chargement
+# ---------------------------------------------------------------------------
+
+def load_benchmark_json(path: str | Path) -> dict[str, Any]:
+ """Charge un JSON de benchmark depuis disque.
+
+ Accepte :
+ - le format ``BenchmarkResult.as_dict()`` (clé ``ranking``,
+ ``engine_reports`` ou ``engines``) ;
+ - un dict déjà parsé ; dans ce cas, ``path`` peut être un dict.
+ """
+ if isinstance(path, dict):
+ return path
+ p = Path(path)
+ if not p.exists():
+ raise FileNotFoundError(f"Fichier benchmark introuvable : {p}")
+ with p.open(encoding="utf-8") as fh:
+ data = json.load(fh)
+ if not isinstance(data, dict):
+ raise ValueError(f"Le JSON {p} doit être un dict.")
+ return data
+
+
+# ---------------------------------------------------------------------------
+# Comparaison
+# ---------------------------------------------------------------------------
+
+def _ranking_index(data: dict[str, Any]) -> dict[str, dict[str, Any]]:
+ """Indexe ``ranking`` par nom de moteur — robuste aux deux formats.
+
+ Un ``BenchmarkResult.as_dict()`` expose ``ranking`` directement
+ (clés ``engine``, ``mean_cer``, …). Le format alternatif ``engines``
+ expose le même contenu sous des clés légèrement différentes —
+ on normalise vers le format ``ranking``.
+ """
+ ranking = data.get("ranking")
+ if isinstance(ranking, list) and ranking:
+ return {
+ r["engine"]: {
+ "engine": r["engine"],
+ "mean_cer": r.get("mean_cer"),
+ "mean_wer": r.get("mean_wer"),
+ "documents": int(r.get("documents") or 0),
+ "failed": int(r.get("failed") or 0),
+ }
+ for r in ranking
+ if isinstance(r, dict) and r.get("engine")
+ }
+ # Fallback : ``engines`` (format report_data)
+ engines = data.get("engines") or []
+ out: dict[str, dict[str, Any]] = {}
+ if isinstance(engines, list):
+ for e in engines:
+ if not isinstance(e, dict):
+ continue
+ name = e.get("name") or e.get("engine")
+ if not name:
+ continue
+ out[name] = {
+ "engine": name,
+ "mean_cer": e.get("cer"),
+ "mean_wer": e.get("wer"),
+ "documents": int(e.get("documents") or 0),
+ "failed": int(e.get("failed") or 0),
+ }
+ return out
+
+
+def _label_of(data: dict[str, Any], default: str) -> str:
+ meta = data.get("meta") or {}
+ return (
+ meta.get("corpus_name")
+ or (data.get("corpus") or {}).get("name")
+ or default
+ )
+
+
+def _run_date_of(data: dict[str, Any]) -> Optional[str]:
+ return (
+ data.get("run_date")
+ or (data.get("meta") or {}).get("run_date")
+ )
+
+
+def _corpus_of(data: dict[str, Any]) -> Optional[str]:
+ meta = data.get("meta") or {}
+ return (
+ meta.get("corpus_source")
+ or (data.get("corpus") or {}).get("source")
+ or meta.get("corpus_name")
+ )
+
+
+def _safe_delta(a: Optional[float], b: Optional[float]) -> Optional[float]:
+ if a is None or b is None:
+ return None
+ return float(b) - float(a)
+
+
+def compare_benchmarks(
+ a: str | Path | dict[str, Any],
+ b: str | Path | dict[str, Any],
+ *,
+ threshold: float = 0.005,
+ label_a: str = "A",
+ label_b: str = "B",
+) -> ComparisonResult:
+ """Compare deux runs et retourne les deltas par moteur.
+
+ Convention : un delta CER positif signifie que ``b`` est *moins bon*
+ que ``a`` (régression). Un seuil ``threshold`` strictement positif
+ (en fraction, ex. 0,005 = 0,5 pp) discrimine régression / bruit.
+ """
+ da = load_benchmark_json(a) if not isinstance(a, dict) else a
+ db = load_benchmark_json(b) if not isinstance(b, dict) else b
+
+ idx_a = _ranking_index(da)
+ idx_b = _ranking_index(db)
+
+ common = sorted(set(idx_a) & set(idx_b))
+ only_a = sorted(set(idx_a) - set(idx_b))
+ only_b = sorted(set(idx_b) - set(idx_a))
+
+ deltas: list[EngineDelta] = []
+ for name in common:
+ ea = idx_a[name]
+ eb = idx_b[name]
+ delta_cer = _safe_delta(ea["mean_cer"], eb["mean_cer"])
+ delta_wer = _safe_delta(ea["mean_wer"], eb["mean_wer"])
+ regression = bool(delta_cer is not None and delta_cer > threshold)
+ improvement = bool(delta_cer is not None and delta_cer < -threshold)
+ deltas.append(
+ EngineDelta(
+ engine=name,
+ cer_a=ea["mean_cer"],
+ cer_b=eb["mean_cer"],
+ delta_cer=delta_cer,
+ wer_a=ea["mean_wer"],
+ wer_b=eb["mean_wer"],
+ delta_wer=delta_wer,
+ docs_a=int(ea["documents"]),
+ docs_b=int(eb["documents"]),
+ failed_a=int(ea["failed"]),
+ failed_b=int(eb["failed"]),
+ is_regression=regression,
+ is_improvement=improvement,
+ )
+ )
+
+ # Tri : régressions (delta décroissant) puis améliorations (delta croissant).
+ deltas.sort(key=lambda d: (
+ not d.is_regression,
+ -(d.delta_cer if d.delta_cer is not None else 0.0),
+ ))
+
+ return ComparisonResult(
+ label_a=label_a,
+ label_b=label_b,
+ run_date_a=_run_date_of(da),
+ run_date_b=_run_date_of(db),
+ corpus_a=_corpus_of(da),
+ corpus_b=_corpus_of(db),
+ deltas=deltas,
+ only_in_a=only_a,
+ only_in_b=only_b,
+ threshold=float(threshold),
+ )
+
+
+def detect_regressions(
+ diff: ComparisonResult,
+) -> list[EngineDelta]:
+ """Retourne uniquement les moteurs en régression dans ``diff``."""
+ return [d for d in diff.deltas if d.is_regression]
+
+
+# ---------------------------------------------------------------------------
+# Rendu HTML
+# ---------------------------------------------------------------------------
+
+_COMPARISON_TEMPLATE = """
+
+
+
+
-
-
-
- Classement des moteurs
- -
-
-
-
-
- | #↑ | -Concurrent↕ | -CER exact↕ | -CER diplo.↕ | -WER↕ | -MER↕ | -WIL↕ | -Ligatures↕ | -Diacritiques↕ | -Gini↕ | -Ancrage↕ | -CER médian | -CER min | -CER max | -Sur-norm. | -Docs | -
|---|
-
-
-
- {% if stratified_ranking_html %}
- {{ stratified_ranking_html }}
- {% endif %}
-
- CER < 5 %
-
-
- 5–15 %
-
-
- 15–30 %
-
-
- > 30 %
-
-
-
-Analyse robuste (sans hallucinations)
-- Recalcule CER, WER, MER, WIL, Gini et ancrage en excluant les documents détectés comme hallucinés ou problématiques. - Cochez/décochez des documents dans la Galerie pour les exclure manuellement. -
-
-
-
-
-
-
-
-
- Comparaison : {{ diff.label_a }} → {{ diff.label_b }}
+ + +Moteurs comparés ({{ diff.deltas|length }})
+{% if not diff.deltas %} +Aucun moteur commun aux deux runs.
+{% else %} +| Moteur | +CER A | +CER B | +Δ CER | +Docs A → B | +État | +
|---|---|---|---|---|---|
| {{ d.engine }} | +{{ "%.3f"|format(d.cer_a) if d.cer_a is not none else "—" }} | +{{ "%.3f"|format(d.cer_b) if d.cer_b is not none else "—" }} | ++ {% if d.delta_cer is none %}— + {% elif d.delta_cer > 0 %}+{{ "%.3f"|format(d.delta_cer) }} + {% else %}{{ "%.3f"|format(d.delta_cer) }} + {% endif %} + | +{{ d.docs_a }} → {{ d.docs_b }} | ++ {% if d.is_regression %}régression + {% elif d.is_improvement %}amélioration + {% else %}{% endif %} + | +
Présents uniquement dans A
+- {% for n in diff.only_in_a %}
- {{ n }} {% endfor %}
Présents uniquement dans B
+- {% for n in diff.only_in_b %}
- {{ n }} {% endfor %}
- \s*
- ', html) def test_report_synthesis_is_deterministic(self, benchmark_result, tmp_path): - from picarones.report.generator import ReportGenerator + from picarones.reports_v2.html.generator import ReportGenerator out1 = tmp_path / "r1.html" out2 = tmp_path / "r2.html" ReportGenerator(benchmark_result).generate(out1) @@ -554,7 +554,7 @@ class TestReportIntegration: assert registered == set(FactType) def test_english_locale_produces_english_sentences(self, benchmark_result, tmp_path): - from picarones.report.generator import ReportGenerator + from picarones.reports_v2.html.generator import ReportGenerator out = tmp_path / "report_en.html" ReportGenerator(benchmark_result, lang="en").generate(out) html = out.read_text(encoding="utf-8") diff --git a/tests/measurements/test_sprint20_pareto_pricing.py b/tests/measurements/test_sprint20_pareto_pricing.py index bd80d6f09262ba4a4ad8ee4d11573b86433f6501..00139dde33a3fe610cad126dc47565c513e86b5d 100644 --- a/tests/measurements/test_sprint20_pareto_pricing.py +++ b/tests/measurements/test_sprint20_pareto_pricing.py @@ -267,7 +267,7 @@ def benchmark_result(): class TestReportIntegration: def test_report_contains_pareto_card(self, benchmark_result, tmp_path): - from picarones.report.generator import ReportGenerator + from picarones.reports_v2.html.generator import ReportGenerator out = tmp_path / "report.html" ReportGenerator(benchmark_result).generate(out) html = out.read_text(encoding="utf-8") @@ -279,7 +279,7 @@ class TestReportIntegration: assert "pareto-experimental" in html # étiquette expérimental def test_report_json_contains_pareto_data(self, benchmark_result): - from picarones.report.generator import _build_report_data + from picarones.reports_v2.html.generator import _build_report_data data = _build_report_data(benchmark_result, images_b64={}) pareto = data.get("pareto", {}) assert "cost" in pareto @@ -293,7 +293,7 @@ class TestReportIntegration: def test_synthesis_may_include_pareto_sentence(self, benchmark_result, tmp_path): # Sur la fixture de démo, pero_ocr + tesseract sont sur le front → la # synthèse doit remonter une alternative moins chère - from picarones.report.generator import ReportGenerator + from picarones.reports_v2.html.generator import ReportGenerator out = tmp_path / "report.html" ReportGenerator(benchmark_result).generate(out) html = out.read_text(encoding="utf-8") @@ -310,7 +310,7 @@ class TestReportIntegration: assert Path(_DEFAULT_PRICING_PATH).exists() def test_english_locale_renders_pareto_labels(self, benchmark_result, tmp_path): - from picarones.report.generator import ReportGenerator + from picarones.reports_v2.html.generator import ReportGenerator out = tmp_path / "report_en.html" ReportGenerator(benchmark_result, lang="en").generate(out) html = out.read_text(encoding="utf-8") diff --git a/tests/report/test_a11y_level_a.py b/tests/report/test_a11y_level_a.py index a634f50167a8c2b1f04fe8fc35861c1bf9b54a35..da6764b91c89d8a592c538f70f5f2c5451235aef 100644 --- a/tests/report/test_a11y_level_a.py +++ b/tests/report/test_a11y_level_a.py @@ -25,7 +25,7 @@ import re import pytest from picarones.fixtures import generate_sample_benchmark -from picarones.report.generator import ReportGenerator +from picarones.reports_v2.html.generator import ReportGenerator @pytest.fixture(scope="module") diff --git a/tests/report/test_a11y_level_aa.py b/tests/report/test_a11y_level_aa.py index b28bcfcaecf1a5ea755ee2f29c825d10c57d5bfc..68116e964a623344ddabd3b230c2b08165576171 100644 --- a/tests/report/test_a11y_level_aa.py +++ b/tests/report/test_a11y_level_aa.py @@ -20,7 +20,7 @@ import re import pytest from picarones.fixtures import generate_sample_benchmark -from picarones.report.generator import ReportGenerator +from picarones.reports_v2.html.generator import ReportGenerator @pytest.fixture(scope="module") diff --git a/tests/report/test_extra_metrics.py b/tests/report/test_extra_metrics.py index 8b5daa54a8f1014e9bcca6c10abcfd5304bb0117..dd64379d5de3e662b62c2d49d1c2926a8bb8f0d7 100644 --- a/tests/report/test_extra_metrics.py +++ b/tests/report/test_extra_metrics.py @@ -16,8 +16,8 @@ from __future__ import annotations import pytest from picarones.fixtures import generate_sample_benchmark -from picarones.report.report_data import build_report_data -from picarones.report.report_data.extra_metrics import ( +from picarones.reports_v2.html.data import build_report_data +from picarones.reports_v2.html.data.extra_metrics import ( compute_marginal_cost_section, compute_rare_token_recall_per_engine, compute_taxonomy_cooccurrence_section, diff --git a/tests/report/test_lazy_images.py b/tests/report/test_lazy_images.py index 398a7e758e519b84d0f724ac61cb5bb206dcb0a6..bbe559d329035d379907e4ec87e143577394d67c 100644 --- a/tests/report/test_lazy_images.py +++ b/tests/report/test_lazy_images.py @@ -55,7 +55,7 @@ def demo_benchmark_with_images(tmp_path: Path): def test_default_mode_inlines_images(demo_benchmark_with_images, tmp_path: Path) -> None: """``lazy_images=False`` (défaut) : les images vivent en base64 inline dans le HTML, aucun fichier d'asset n'est créé.""" - from picarones.report.generator import ReportGenerator + from picarones.reports_v2.html.generator import ReportGenerator out = tmp_path / "report.html" gen = ReportGenerator(demo_benchmark_with_images) @@ -83,7 +83,7 @@ def test_lazy_mode_creates_asset_directory( ) -> None: """``lazy_images=True`` : ``report-assets/`` est créé à côté du HTML et contient des fichiers image.""" - from picarones.report.generator import ReportGenerator + from picarones.reports_v2.html.generator import ReportGenerator out = tmp_path / "report.html" gen = ReportGenerator(demo_benchmark_with_images, lazy_images=True) @@ -104,7 +104,7 @@ def test_lazy_mode_html_references_relative_urls( ) -> None: """En mode lazy, le HTML référence les images via URL relative ``report-assets/...`` plutôt qu'un data-URI.""" - from picarones.report.generator import ReportGenerator + from picarones.reports_v2.html.generator import ReportGenerator out = tmp_path / "report.html" gen = ReportGenerator(demo_benchmark_with_images, lazy_images=True) @@ -127,7 +127,7 @@ def test_lazy_mode_significantly_reduces_html_size( favorable au lazy. Test peu strict (ratio > 1.05) pour ne pas être flaky en fonction du contenu vendor. """ - from picarones.report.generator import ReportGenerator + from picarones.reports_v2.html.generator import ReportGenerator inline_out = tmp_path / "inline.html" lazy_out = tmp_path / "lazy.html" @@ -152,7 +152,7 @@ def test_lazy_mode_with_missing_image_does_not_crash(tmp_path: Path) -> None: """Si l'image source n'existe pas, l'externalisation log un warning et continue (rétrocompat avec ``_encode_image_b64`` qui retourne '' silencieusement).""" - from picarones.report.generator import ReportGenerator + from picarones.reports_v2.html.generator import ReportGenerator bench = generate_sample_benchmark(n_docs=2) # Pointe vers un chemin inexistant @@ -171,7 +171,7 @@ def test_safe_filename_generation(tmp_path: Path) -> None: des noms de fichiers normalisés (pas de path traversal possible).""" from PIL import Image - from picarones.report.generator import _externalize_images_to_dir + from picarones.reports_v2.html.generator import _externalize_images_to_dir src = tmp_path / "src.png" Image.new("RGB", (50, 50), color=(0, 0, 0)).save(src) diff --git a/tests/report/test_report.py b/tests/report/test_report.py index dcab4e15c75db5a22e769b932fe6157aee7c8863..e401361ca5e0a6db01be6ede77289d447d86f0df 100644 --- a/tests/report/test_report.py +++ b/tests/report/test_report.py @@ -4,7 +4,7 @@ import json import pytest from picarones.fixtures import generate_sample_benchmark -from picarones.report.generator import ReportGenerator, _build_report_data, _cer_color +from picarones.reports_v2.html.generator import ReportGenerator, _build_report_data, _cer_color # --------------------------------------------------------------------------- diff --git a/tests/report/test_sprint17_jinja2_refactor.py b/tests/report/test_sprint17_jinja2_refactor.py index a71d9c310fb0ce6ddbf1c93f647c0cec25c8ce79..3a044651a9adb67b2e92c5912d7502bed2d43600 100644 --- a/tests/report/test_sprint17_jinja2_refactor.py +++ b/tests/report/test_sprint17_jinja2_refactor.py @@ -16,7 +16,7 @@ from pathlib import Path import pytest from picarones import fixtures -from picarones.report.generator import ( +from picarones.reports_v2.html.generator import ( ReportGenerator, _build_jinja_env, _TEMPLATES_DIR, @@ -153,7 +153,7 @@ class TestI18nFromJSON: def test_fr_and_en_have_same_keys(self): """Garde-fou contre les traductions manquantes.""" - from picarones.i18n import TRANSLATIONS + from picarones.reports_v2.i18n import TRANSLATIONS fr_keys = set(TRANSLATIONS.get("fr", {}).keys()) en_keys = set(TRANSLATIONS.get("en", {}).keys()) missing_in_en = fr_keys - en_keys @@ -162,7 +162,7 @@ class TestI18nFromJSON: assert not missing_in_fr, f"Clés manquantes en français : {missing_in_fr}" def test_translations_load_via_public_api(self): - from picarones.i18n import get_labels, SUPPORTED_LANGS + from picarones.reports_v2.i18n import get_labels, SUPPORTED_LANGS assert "fr" in SUPPORTED_LANGS assert "en" in SUPPORTED_LANGS fr = get_labels("fr") diff --git a/tests/report/test_sprint21_glossary_customize.py b/tests/report/test_sprint21_glossary_customize.py index e0d04f4c552238ce8134ca9ea7ea81ed46da2c66..5cebbd7bb447e107b12f36b047adcdaf106b9631 100644 --- a/tests/report/test_sprint21_glossary_customize.py +++ b/tests/report/test_sprint21_glossary_customize.py @@ -131,7 +131,7 @@ def benchmark_result(): class TestReportIntegration: def test_report_embeds_glossary_json(self, benchmark_result, tmp_path): - from picarones.report.generator import ReportGenerator + from picarones.reports_v2.html.generator import ReportGenerator out = tmp_path / "report.html" ReportGenerator(benchmark_result).generate(out) html = out.read_text(encoding="utf-8") @@ -142,7 +142,7 @@ class TestReportIntegration: assert "definition" in glossary["cer"] def test_report_contains_side_panels(self, benchmark_result, tmp_path): - from picarones.report.generator import ReportGenerator + from picarones.reports_v2.html.generator import ReportGenerator out = tmp_path / "report.html" ReportGenerator(benchmark_result).generate(out) html = out.read_text(encoding="utf-8") @@ -151,7 +151,7 @@ class TestReportIntegration: assert 'class="side-panel-close"' in html def test_report_has_advanced_button_in_nav(self, benchmark_result, tmp_path): - from picarones.report.generator import ReportGenerator + from picarones.reports_v2.html.generator import ReportGenerator out = tmp_path / "report.html" ReportGenerator(benchmark_result).generate(out) html = out.read_text(encoding="utf-8") @@ -159,7 +159,7 @@ class TestReportIntegration: assert 'openCustomize()' in html def test_ranking_columns_have_glossary_keys(self, benchmark_result, tmp_path): - from picarones.report.generator import ReportGenerator + from picarones.reports_v2.html.generator import ReportGenerator out = tmp_path / "report.html" ReportGenerator(benchmark_result).generate(out) html = out.read_text(encoding="utf-8") @@ -167,7 +167,7 @@ class TestReportIntegration: assert f'data-glossary-key="{k}"' in html, f"Header pour {k} sans data-glossary-key" def test_app_js_has_glossary_and_customize_functions(self, benchmark_result, tmp_path): - from picarones.report.generator import ReportGenerator + from picarones.reports_v2.html.generator import ReportGenerator out = tmp_path / "report.html" ReportGenerator(benchmark_result).generate(out) html = out.read_text(encoding="utf-8") @@ -182,7 +182,7 @@ class TestReportIntegration: assert fn in html, f"Fonction {fn} manquante" def test_english_glossary_for_en_locale(self, benchmark_result, tmp_path): - from picarones.report.generator import ReportGenerator + from picarones.reports_v2.html.generator import ReportGenerator out = tmp_path / "report_en.html" ReportGenerator(benchmark_result, lang="en").generate(out) html = out.read_text(encoding="utf-8") @@ -198,7 +198,7 @@ class TestReportIntegration: class TestI18nKeysForCustomize: def test_required_customize_keys_present(self): - from picarones.i18n import get_labels + from picarones.reports_v2.i18n import get_labels required = { "btn_customize", "customize_title", "customize_columns", "customize_filters", @@ -223,7 +223,7 @@ class TestNoPrescriptionGuards: et que les poids de score composite sont à 0 par défaut côté JS.""" def test_warning_message_is_visible(self, tmp_path): - from picarones.report.generator import ReportGenerator + from picarones.reports_v2.html.generator import ReportGenerator from picarones import fixtures bench = fixtures.generate_sample_benchmark(n_docs=3) out = tmp_path / "r.html" @@ -234,7 +234,7 @@ class TestNoPrescriptionGuards: "universally valid weighting" in html def test_default_weights_are_empty_in_js_state(self, tmp_path): - from picarones.report.generator import ReportGenerator + from picarones.reports_v2.html.generator import ReportGenerator from picarones import fixtures bench = fixtures.generate_sample_benchmark(n_docs=3) out = tmp_path / "r.html" diff --git a/tests/report/test_sprint22_docs_case_studies.py b/tests/report/test_sprint22_docs_case_studies.py index 5a594881b71d4122822667196540682ee75c356c..ffbfee2046820d6fa0d242ace04f03a41f0f8050 100644 --- a/tests/report/test_sprint22_docs_case_studies.py +++ b/tests/report/test_sprint22_docs_case_studies.py @@ -116,7 +116,7 @@ def benchmark_result(): class TestReportIntegration: def test_report_links_to_case_studies(self, benchmark_result, tmp_path): - from picarones.report.generator import ReportGenerator + from picarones.reports_v2.html.generator import ReportGenerator out = tmp_path / "report.html" ReportGenerator(benchmark_result).generate(out) html = out.read_text(encoding="utf-8") @@ -125,7 +125,7 @@ class TestReportIntegration: def test_report_polish_no_consecutive_empty_lines_in_views(self, benchmark_result, tmp_path): """Garde-fou cosmétique léger — éviter les blocs vides excessifs introduits par les includes Jinja2.""" - from picarones.report.generator import ReportGenerator + from picarones.reports_v2.html.generator import ReportGenerator out = tmp_path / "report.html" ReportGenerator(benchmark_result).generate(out) html = out.read_text(encoding="utf-8") @@ -143,7 +143,7 @@ class TestEndToEnd: def test_small_corpus_renders(self, tmp_path): from picarones import fixtures - from picarones.report.generator import ReportGenerator + from picarones.reports_v2.html.generator import ReportGenerator bench = fixtures.generate_sample_benchmark(n_docs=2) out = tmp_path / "small.html" ReportGenerator(bench).generate(out) @@ -151,7 +151,7 @@ class TestEndToEnd: def test_large_corpus_renders(self, tmp_path): from picarones import fixtures - from picarones.report.generator import ReportGenerator + from picarones.reports_v2.html.generator import ReportGenerator bench = fixtures.generate_sample_benchmark(n_docs=20) out = tmp_path / "large.html" ReportGenerator(bench).generate(out) @@ -159,7 +159,7 @@ class TestEndToEnd: def test_english_locale_full_render(self, tmp_path): from picarones import fixtures - from picarones.report.generator import ReportGenerator + from picarones.reports_v2.html.generator import ReportGenerator bench = fixtures.generate_sample_benchmark(n_docs=5) out = tmp_path / "en.html" ReportGenerator(bench, lang="en").generate(out) diff --git a/tests/report/test_sprint27_reproducibility_snapshots.py b/tests/report/test_sprint27_reproducibility_snapshots.py index 2018c1dedcbf96e41782b8285761666e47a3cf49..9568c0c8895f2fd3a52e4b531161cf17024e9f17 100644 --- a/tests/report/test_sprint27_reproducibility_snapshots.py +++ b/tests/report/test_sprint27_reproducibility_snapshots.py @@ -28,7 +28,7 @@ import pytest class TestPricingSnapshot: def test_default_pricing_yaml_is_loaded(self): - from picarones.report.snapshot import pricing_snapshot + from picarones.reports_v2.html.snapshot import pricing_snapshot s = pricing_snapshot() assert s["available"] is True assert s["filename"] == "pricing.yaml" @@ -38,19 +38,19 @@ class TestPricingSnapshot: assert isinstance(s["data"], dict) def test_data_contains_meta_and_engines(self): - from picarones.report.snapshot import pricing_snapshot + from picarones.reports_v2.html.snapshot import pricing_snapshot s = pricing_snapshot() assert "meta" in s["data"], "le snapshot doit exposer la section meta" assert "engines" in s["data"], "le snapshot doit exposer engines" def test_missing_path_returns_unavailable(self, tmp_path): - from picarones.report.snapshot import pricing_snapshot + from picarones.reports_v2.html.snapshot import pricing_snapshot s = pricing_snapshot(pricing_path=tmp_path / "ne-pas-exister.yaml") assert s["available"] is False assert "introuvable" in s["reason"].lower() def test_custom_yaml_round_trips(self, tmp_path): - from picarones.report.snapshot import pricing_snapshot + from picarones.reports_v2.html.snapshot import pricing_snapshot custom = tmp_path / "custom.yaml" custom.write_text( "meta:\n currency: USD\n last_updated: 2026-01-01\nengines:\n fake: {type: local}\n", @@ -66,7 +66,7 @@ class TestPricingSnapshot: class TestGlossarySnapshot: def test_default_lang_returns_entries(self): - from picarones.report.snapshot import glossary_snapshot + from picarones.reports_v2.html.snapshot import glossary_snapshot s = glossary_snapshot(lang="fr") assert s["available"] is True assert s["entry_count"] > 10 @@ -75,7 +75,7 @@ class TestGlossarySnapshot: assert k in s["entries"] def test_used_keys_filter(self): - from picarones.report.snapshot import glossary_snapshot + from picarones.reports_v2.html.snapshot import glossary_snapshot s = glossary_snapshot(lang="fr", used_keys=["cer"]) assert s["entry_count"] == 1 assert list(s["entries"]) == ["cer"] @@ -84,13 +84,13 @@ class TestGlossarySnapshot: # `load_glossary` retombe sur fr si la langue est absente — donc # le snapshot doit être disponible avec lang='fr' ou la langue # demandée selon ce qu'on retourne. On vérifie qu'on ne crashe pas. - from picarones.report.snapshot import glossary_snapshot + from picarones.reports_v2.html.snapshot import glossary_snapshot s = glossary_snapshot(lang="xx-pas-existante") # Soit on retombe sur fr (available=True), soit on signale unavailable. assert "available" in s def test_entries_sorted_for_determinism(self): - from picarones.report.snapshot import glossary_snapshot + from picarones.reports_v2.html.snapshot import glossary_snapshot s = glossary_snapshot(lang="fr") keys = list(s["entries"]) assert keys == sorted(keys), ( @@ -102,7 +102,7 @@ class TestGlossarySnapshot: class TestNormalizationSnapshot: def test_builtin_profile_serializes(self): from picarones.measurements.normalization import get_builtin_profile - from picarones.report.snapshot import normalization_snapshot + from picarones.reports_v2.html.snapshot import normalization_snapshot p = get_builtin_profile("medieval_french") s = normalization_snapshot(p) assert s["available"] is True @@ -112,13 +112,13 @@ class TestNormalizationSnapshot: assert s["diplomatic_table"].get("ſ") == "s" def test_none_profile_returns_unavailable(self): - from picarones.report.snapshot import normalization_snapshot + from picarones.reports_v2.html.snapshot import normalization_snapshot s = normalization_snapshot(None) assert s["available"] is False def test_exclude_chars_sorted(self): from picarones.measurements.normalization import get_builtin_profile - from picarones.report.snapshot import normalization_snapshot + from picarones.reports_v2.html.snapshot import normalization_snapshot p = get_builtin_profile("sans_ponctuation") s = normalization_snapshot(p) # Liste triée pour reproductibilité @@ -128,20 +128,20 @@ class TestNormalizationSnapshot: class TestEnvironmentSnapshot: def test_returns_picarones_version(self): from picarones import __version__ - from picarones.report.snapshot import environment_snapshot + from picarones.reports_v2.html.snapshot import environment_snapshot s = environment_snapshot() assert s["available"] is True assert s["picarones_version"] == __version__ def test_python_and_platform_present(self): - from picarones.report.snapshot import environment_snapshot + from picarones.reports_v2.html.snapshot import environment_snapshot s = environment_snapshot() assert s["python_version"] assert s["python_implementation"] assert s["platform"] def test_installed_packages_sorted_unique(self): - from picarones.report.snapshot import environment_snapshot + from picarones.reports_v2.html.snapshot import environment_snapshot s = environment_snapshot() pkgs = s["installed_packages"] assert isinstance(pkgs, list) @@ -152,7 +152,7 @@ class TestEnvironmentSnapshot: assert len(names) == len(set(names)) def test_git_commit_is_str_or_none(self): - from picarones.report.snapshot import environment_snapshot + from picarones.reports_v2.html.snapshot import environment_snapshot s = environment_snapshot() commit = s.get("git_commit") assert commit is None or (isinstance(commit, str) and 0 < len(commit) <= 12) @@ -164,7 +164,7 @@ class TestEnvironmentSnapshot: class TestSnapshotAll: def test_contains_all_four_blocks(self): - from picarones.report.snapshot import snapshot_all + from picarones.reports_v2.html.snapshot import snapshot_all s = snapshot_all() for k in ("pricing", "glossary", "normalization", "environment"): assert k in s, f"snapshot_all doit exposer la clé '{k}'" @@ -172,7 +172,7 @@ class TestSnapshotAll: def test_deterministic_for_same_inputs(self): from picarones.measurements.normalization import get_builtin_profile - from picarones.report.snapshot import snapshot_all + from picarones.reports_v2.html.snapshot import snapshot_all profile = get_builtin_profile("nfc") a = snapshot_all(lang="fr", normalization_profile=profile) @@ -193,7 +193,7 @@ def generated_report_html(tmp_path_factory) -> str: """Génère un rapport démo et retourne son contenu HTML.""" from picarones import fixtures from picarones.measurements.normalization import get_builtin_profile - from picarones.report.generator import ReportGenerator + from picarones.reports_v2.html.generator import ReportGenerator b = fixtures.generate_sample_benchmark(n_docs=6) out_dir = tmp_path_factory.mktemp("rep27") diff --git a/tests/report/test_sprint37_inter_engine_html.py b/tests/report/test_sprint37_inter_engine_html.py index 8a312bbc3b28b8fd61bb52d8308c31a2b4898d82..3632a2c8bcf24ee581365ff883667646ed399b6a 100644 --- a/tests/report/test_sprint37_inter_engine_html.py +++ b/tests/report/test_sprint37_inter_engine_html.py @@ -26,7 +26,7 @@ from pathlib import Path import pytest from picarones.fixtures import generate_sample_benchmark -from picarones.report.generator import ReportGenerator +from picarones.reports_v2.html.generator import ReportGenerator from picarones.reports_v2.html.renderers.inter_engine import ( build_divergence_matrix_html, build_oracle_gap_html, diff --git a/tests/report/test_sprint41_ner_html.py b/tests/report/test_sprint41_ner_html.py index f5a30715df8ff96bb7dd722f4ae2a82cd44629c0..ca19c95f1ece0ceddd95c797014e5c71e15cc133 100644 --- a/tests/report/test_sprint41_ner_html.py +++ b/tests/report/test_sprint41_ner_html.py @@ -22,7 +22,7 @@ from pathlib import Path import pytest from picarones.fixtures import generate_sample_benchmark -from picarones.report.generator import ReportGenerator +from picarones.reports_v2.html.generator import ReportGenerator from picarones.reports_v2.html.renderers.ner import ( build_ner_per_category_html, build_ner_summary_html, diff --git a/tests/report/test_sprint43_calibration_html.py b/tests/report/test_sprint43_calibration_html.py index 6a8f58e43d1e60b9bebd95b7d052ce6952ef7758..78049cbe414f12254882e855621f83072dd8b482 100644 --- a/tests/report/test_sprint43_calibration_html.py +++ b/tests/report/test_sprint43_calibration_html.py @@ -31,7 +31,7 @@ from picarones.reports_v2.html.renderers.calibration import ( build_reliability_diagram_svg, build_reliability_diagrams_grid_html, ) -from picarones.report.generator import ReportGenerator +from picarones.reports_v2.html.generator import ReportGenerator # ────────────────────────────────────────────────────────────────────────── diff --git a/tests/report/test_sprint46_stratification_html.py b/tests/report/test_sprint46_stratification_html.py index 006fff05b0f6dbec405027d04cf539d8912d497c..24bf5493b6f23d036b9e1075946a672521cd480d 100644 --- a/tests/report/test_sprint46_stratification_html.py +++ b/tests/report/test_sprint46_stratification_html.py @@ -31,7 +31,7 @@ from picarones.measurements.narrative.detectors import detect_stratification_rec from picarones.core.facts import FactImportance, FactType from picarones.measurements.narrative.renderer import extract_numbers, render_fact from picarones.core.results import DocumentResult -from picarones.report.generator import ReportGenerator +from picarones.reports_v2.html.generator import ReportGenerator from picarones.reports_v2.html.renderers.stratification import build_stratified_ranking_html diff --git a/tests/report/test_sprint5_advanced_metrics.py b/tests/report/test_sprint5_advanced_metrics.py index 4c1b7b2d4fa910567eaa0cc4c81ac5e900e483ba..1cf92cbcfbb9c3edef072d68e8817628cd211ea5 100644 --- a/tests/report/test_sprint5_advanced_metrics.py +++ b/tests/report/test_sprint5_advanced_metrics.py @@ -759,7 +759,7 @@ class TestReportSprint5: def test_report_data_has_ligature_score(self): from picarones.fixtures import generate_sample_benchmark - from picarones.report.generator import _build_report_data + from picarones.reports_v2.html.generator import _build_report_data bm = generate_sample_benchmark() data = _build_report_data(bm, {}) for eng in data["engines"]: @@ -767,7 +767,7 @@ class TestReportSprint5: def test_report_data_has_diacritic_score(self): from picarones.fixtures import generate_sample_benchmark - from picarones.report.generator import _build_report_data + from picarones.reports_v2.html.generator import _build_report_data bm = generate_sample_benchmark() data = _build_report_data(bm, {}) for eng in data["engines"]: @@ -775,7 +775,7 @@ class TestReportSprint5: def test_report_data_has_aggregated_taxonomy(self): from picarones.fixtures import generate_sample_benchmark - from picarones.report.generator import _build_report_data + from picarones.reports_v2.html.generator import _build_report_data bm = generate_sample_benchmark() data = _build_report_data(bm, {}) for eng in data["engines"]: @@ -783,7 +783,7 @@ class TestReportSprint5: def test_report_data_has_aggregated_image_quality(self): from picarones.fixtures import generate_sample_benchmark - from picarones.report.generator import _build_report_data + from picarones.reports_v2.html.generator import _build_report_data bm = generate_sample_benchmark() data = _build_report_data(bm, {}) for eng in data["engines"]: @@ -791,7 +791,7 @@ class TestReportSprint5: def test_html_has_characters_tab(self, tmp_path): from picarones.fixtures import generate_sample_benchmark - from picarones.report.generator import ReportGenerator + from picarones.reports_v2.html.generator import ReportGenerator bm = generate_sample_benchmark() out = tmp_path / "report.html" ReportGenerator(bm).generate(out) @@ -800,7 +800,7 @@ class TestReportSprint5: def test_html_has_ligatures_column(self, tmp_path): from picarones.fixtures import generate_sample_benchmark - from picarones.report.generator import ReportGenerator + from picarones.reports_v2.html.generator import ReportGenerator bm = generate_sample_benchmark() out = tmp_path / "report.html" ReportGenerator(bm).generate(out) @@ -809,7 +809,7 @@ class TestReportSprint5: def test_html_has_diacritiques_column(self, tmp_path): from picarones.fixtures import generate_sample_benchmark - from picarones.report.generator import ReportGenerator + from picarones.reports_v2.html.generator import ReportGenerator bm = generate_sample_benchmark() out = tmp_path / "report.html" ReportGenerator(bm).generate(out) @@ -818,7 +818,7 @@ class TestReportSprint5: def test_html_has_scatter_plot(self, tmp_path): from picarones.fixtures import generate_sample_benchmark - from picarones.report.generator import ReportGenerator + from picarones.reports_v2.html.generator import ReportGenerator bm = generate_sample_benchmark() out = tmp_path / "report.html" ReportGenerator(bm).generate(out) @@ -827,7 +827,7 @@ class TestReportSprint5: def test_html_has_taxonomy_chart(self, tmp_path): from picarones.fixtures import generate_sample_benchmark - from picarones.report.generator import ReportGenerator + from picarones.reports_v2.html.generator import ReportGenerator bm = generate_sample_benchmark() out = tmp_path / "report.html" ReportGenerator(bm).generate(out) @@ -836,7 +836,7 @@ class TestReportSprint5: def test_html_has_confusion_heatmap(self, tmp_path): from picarones.fixtures import generate_sample_benchmark - from picarones.report.generator import ReportGenerator + from picarones.reports_v2.html.generator import ReportGenerator bm = generate_sample_benchmark() out = tmp_path / "report.html" ReportGenerator(bm).generate(out) @@ -845,7 +845,7 @@ class TestReportSprint5: def test_doc_results_have_image_quality_in_report(self): from picarones.fixtures import generate_sample_benchmark - from picarones.report.generator import _build_report_data + from picarones.reports_v2.html.generator import _build_report_data bm = generate_sample_benchmark() data = _build_report_data(bm, {}) doc = data["documents"][0] diff --git a/tests/report/test_sprint7_advanced_report.py b/tests/report/test_sprint7_advanced_report.py index cb2c1c284a71eb0af1ede1a4d88ee936e659f637..a23900ebc3ff33f02e45989a70acd9677de5e035 100644 --- a/tests/report/test_sprint7_advanced_report.py +++ b/tests/report/test_sprint7_advanced_report.py @@ -31,14 +31,14 @@ def sample_benchmark_s7(): @pytest.fixture def report_data_s7(sample_benchmark_s7): - from picarones.report.generator import _build_report_data + from picarones.reports_v2.html.generator import _build_report_data imgs = sample_benchmark_s7.metadata.get("_images_b64", {}) return _build_report_data(sample_benchmark_s7, imgs) @pytest.fixture def html_s7(sample_benchmark_s7): - from picarones.report.generator import ReportGenerator + from picarones.reports_v2.html.generator import ReportGenerator import tempfile import pathlib gen = ReportGenerator(sample_benchmark_s7) diff --git a/tests/report/test_views.py b/tests/report/test_views.py index e3f11421eaf08708877627a73877e243faafb29a..8f56aded459b41460ebcd9c1aa3beb50916f7f5d 100644 --- a/tests/report/test_views.py +++ b/tests/report/test_views.py @@ -345,11 +345,11 @@ class TestGeneratorWiring: une docstring) et trop liée à la forme du code. """ from picarones.fixtures import generate_sample_benchmark - from picarones.report.generator import ReportGenerator + from picarones.reports_v2.html.generator import ReportGenerator bench = generate_sample_benchmark() gen = ReportGenerator(bench, lang="fr") - from picarones.i18n import get_labels + from picarones.reports_v2.i18n import get_labels report_data = { "engines": [], @@ -381,7 +381,7 @@ class TestGeneratorWiring: tpl_src = ( Path(__file__).parent.parent.parent - / "picarones" / "report" / "templates" / "view_analyses.html" + / "picarones" / "reports_v2" / "html" / "templates" / "view_analyses.html" ).read_text(encoding="utf-8") assert "{% if economics_view_html %}" in tpl_src assert "{% if advanced_taxonomy_view_html %}" in tpl_src diff --git a/tests/web/test_sprint28_ux_save_compare.py b/tests/web/test_sprint28_ux_save_compare.py index c8125aa78176428e81550a25cbd5f1186f331e0f..3a81338b6ff3a3302f9e6f4afe80e88fbcb62515 100644 --- a/tests/web/test_sprint28_ux_save_compare.py +++ b/tests/web/test_sprint28_ux_save_compare.py @@ -46,7 +46,7 @@ def _benchmark_json(engines_to_cer: dict[str, float], **extra) -> dict: class TestCompareBenchmarks: def test_identical_runs_no_regression(self): - from picarones.report.comparison import compare_benchmarks + from picarones.reports_v2.html.comparison import compare_benchmarks a = _benchmark_json({"tesseract": 0.05, "pero": 0.07}) b = _benchmark_json({"tesseract": 0.05, "pero": 0.07}) diff = compare_benchmarks(a, b, threshold=0.005) @@ -54,7 +54,7 @@ class TestCompareBenchmarks: assert all(not d.is_regression and not d.is_improvement for d in diff.deltas) def test_regression_detected_above_threshold(self): - from picarones.report.comparison import compare_benchmarks, detect_regressions + from picarones.reports_v2.html.comparison import compare_benchmarks, detect_regressions a = _benchmark_json({"tesseract": 0.05}) b = _benchmark_json({"tesseract": 0.06}) # +1 pp diff = compare_benchmarks(a, b, threshold=0.005) @@ -64,7 +64,7 @@ class TestCompareBenchmarks: assert regs[0].delta_cer == pytest.approx(0.01, abs=1e-9) def test_improvement_detected_below_threshold(self): - from picarones.report.comparison import compare_benchmarks + from picarones.reports_v2.html.comparison import compare_benchmarks a = _benchmark_json({"tesseract": 0.05}) b = _benchmark_json({"tesseract": 0.04}) # -1 pp diff = compare_benchmarks(a, b, threshold=0.005) @@ -72,14 +72,14 @@ class TestCompareBenchmarks: assert not diff.deltas[0].is_regression def test_below_threshold_is_stable(self): - from picarones.report.comparison import compare_benchmarks + from picarones.reports_v2.html.comparison import compare_benchmarks a = _benchmark_json({"tesseract": 0.05}) b = _benchmark_json({"tesseract": 0.052}) # +0.2 pp, sous le seuil 0.5 pp diff = compare_benchmarks(a, b, threshold=0.005) assert not diff.deltas[0].is_regression def test_engines_only_in_one_side(self): - from picarones.report.comparison import compare_benchmarks + from picarones.reports_v2.html.comparison import compare_benchmarks a = _benchmark_json({"tesseract": 0.05, "pero": 0.07}) b = _benchmark_json({"tesseract": 0.05, "kraken": 0.06}) diff = compare_benchmarks(a, b, threshold=0.005) @@ -88,7 +88,7 @@ class TestCompareBenchmarks: assert {d.engine for d in diff.deltas} == {"tesseract"} def test_none_cer_does_not_raise(self): - from picarones.report.comparison import compare_benchmarks + from picarones.reports_v2.html.comparison import compare_benchmarks a = _benchmark_json({"tesseract": None}) b = _benchmark_json({"tesseract": 0.05}) diff = compare_benchmarks(a, b) @@ -96,7 +96,7 @@ class TestCompareBenchmarks: assert not diff.deltas[0].is_regression def test_regressions_sorted_by_severity(self): - from picarones.report.comparison import compare_benchmarks + from picarones.reports_v2.html.comparison import compare_benchmarks a = _benchmark_json({"a": 0.05, "b": 0.05, "c": 0.05}) b = _benchmark_json({"a": 0.07, "b": 0.10, "c": 0.06}) # b plus grave diff = compare_benchmarks(a, b, threshold=0.005) @@ -105,7 +105,7 @@ class TestCompareBenchmarks: assert engines_in_order.index("b") < engines_in_order.index("a") def test_loads_from_file_path(self, tmp_path): - from picarones.report.comparison import compare_benchmarks + from picarones.reports_v2.html.comparison import compare_benchmarks a_path = tmp_path / "a.json" b_path = tmp_path / "b.json" a_path.write_text(json.dumps(_benchmark_json({"tesseract": 0.05}))) @@ -116,7 +116,7 @@ class TestCompareBenchmarks: class TestRenderComparisonHTML: def test_html_is_self_contained_and_named(self, tmp_path): - from picarones.report.comparison import compare_benchmarks, render_comparison_html + from picarones.reports_v2.html.comparison import compare_benchmarks, render_comparison_html a = _benchmark_json({"tesseract": 0.05}) b = _benchmark_json({"tesseract": 0.07}) diff = compare_benchmarks(a, b, label_a="V1", label_b="V2")