f70d138db3
Initial commit with full extraction pipeline: PDF OCR (docTR), text segmentation, LLM extraction (Ollama), deterministic post-processing normalizer, validation, and Excel/CSV export. The normalizer fixes OCR/LLM errors on CIM-10 codes: - OCR digit→letter confusion in position 1 (1→I, 0→O, 5→S, 2→Z, 8→B) - Missing dot separator (F050→F05.0, R410→R41.0) - '+' instead of '.' (B99+1→B99.1, J961+0→J96.10) - Excess decimals (Z04.880→Z04.88) - OCR letter→digit in positions 2-3 (LO2.2→L02.2) - Literal "null" string purge - Auto-fill codes_retenus from decision context Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
45 lines
1.2 KiB
Python
45 lines
1.2 KiB
Python
"""
|
|
Configuration T2A Extractor
|
|
"""
|
|
from pathlib import Path
|
|
|
|
# === Ollama ===
|
|
OLLAMA_BASE_URL = "http://localhost:11434"
|
|
OLLAMA_MODEL = "gemma3:27b-cloud" # À adapter selon le tag exact
|
|
OLLAMA_TIMEOUT = 120 # secondes par requête
|
|
OLLAMA_MAX_RETRIES = 2
|
|
|
|
# === OCR (docTR) ===
|
|
DOCTR_DET_ARCH = "db_resnet50"
|
|
DOCTR_RECO_ARCH = "crnn_vgg16_bn"
|
|
OCR_DPI = 200 # résolution pour conversion page → image
|
|
OCR_MIN_CONFIDENCE = 0.5 # seuil de confiance minimum docTR
|
|
|
|
# === Extraction PDF ===
|
|
# Seuil de caractères pour considérer une page comme "native"
|
|
# (certaines pages scannées ont quelques caractères parasites)
|
|
NATIVE_TEXT_MIN_CHARS = 50
|
|
|
|
# === Schéma de sortie ===
|
|
OUTPUT_COLUMNS = [
|
|
"champ",
|
|
"num_ogc",
|
|
"type_desaccord",
|
|
"codes_etablissement",
|
|
"libelle_etablissement",
|
|
"codes_controleurs",
|
|
"libelle_controleurs",
|
|
"decision_ucr",
|
|
"codes_retenus",
|
|
"ghm_ghs",
|
|
"texte_decision",
|
|
]
|
|
|
|
# Valeurs autorisées pour les enums
|
|
DECISION_VALUES = ["Favorable", "Défavorable"]
|
|
TYPE_DESACCORD_VALUES = ["DP", "DAS", "DP+DAS", "Actes"]
|
|
|
|
# === Chemins ===
|
|
PROJECT_ROOT = Path(__file__).parent
|
|
DEFAULT_OUTPUT_DIR = PROJECT_ROOT / "output"
|