Initial commit — Pseudonymisation de PDF v5

- GUI v5 : vue unique épurée (tkinter), 2 étapes visuelles - Core ONNX : anonymisation regex + NER optionnel - Extraction globale des noms depuis champs structurés (Patient, Rédigé par, MME/Madame, DR) - Génération simultanée PDF Image + PDF Anonymisé (structure préservée) - Build Windows via Nuitka (script batch + GitHub Actions CI) - install.sh pour setup/run Linux Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-02-16 15:03:37 +01:00
commit 8339069c83
18 changed files with 5127 additions and 0 deletions
--- a/setup_env_and_build.bat
+++ b/setup_env_and_build.bat
@@ -0,0 +1,216 @@
+@echo off
+setlocal EnableExtensions EnableDelayedExpansion
+
+REM ======== FENETRE PERSISTANTE ========
+if /I not "%~1"=="/keep" (
+  start "" cmd /k "%~f0" /keep
+  goto :eof
+)
+title Setup & Build Pseudonymiseur (Robuste) - PERSISTANT
+
+REM ======== CONFIG ========
+set "PY=py -3.11"
+set "VENV=.venv"
+set "ENTRY=pseudonymisation_pipeline_robuste.py"
+set "EXENAME=PseudonymiseurMedical"
+set "MODEL_DIR=models\fr_core_news_lg"
+set "LOG=build_log.txt"
+set "FR_WHEEL_URL=https://github.com/explosion/spacy-models/releases/download/fr_core_news_lg-3.7.0/fr_core_news_lg-3.7.0-py3-none-any.whl"
+set "SPM_MISSING=1"
+
+REM ======== MENAGE PRECO ========
+echo .
+echo [CLEAN] Nettoyage de l'environnement...
+if exist "Build" del /f /q "Build" >nul 2>&1
+if exist "BUILD" del /f /q "BUILD" >nul 2>&1
+if exist ".\build" rmdir /s /q ".\build" >nul 2>&1
+if exist ".\dist" rmdir /s /q ".\dist" >nul 2>&1
+if exist ".\out"  rmdir /s /q ".\out"  >nul 2>&1
+del /f /q *.spec *.pyc 2>nul
+for /d %%D in (__pycache__ .pytest_cache .mypy_cache) do if exist "%%D" rmdir /s /q "%%D" 2>nul
+echo [CLEAN] OK
+
+echo.
+echo [0] Verif Python 3.11 x64
+%PY% -c "import sys,platform;assert sys.version_info[:2]==(3,11);print(sys.version);print(platform.architecture())"
+if errorlevel 1 (
+  echo [ERREUR] Python 3.11 x64 requis.
+  goto MENU
+)
+
+echo.
+echo [1] Environnement virtuel
+if not exist "%VENV%\Scripts\python.exe" %PY% -m venv "%VENV%"
+if errorlevel 1 (
+  echo [ERREUR] Creation venv impossible.
+  goto MENU
+)
+call "%VENV%\Scripts\activate"
+if errorlevel 1 (
+  echo [ERREUR] Activation venv impossible.
+  goto MENU
+)
+
+echo.
+echo [2] Installation des dependances (voir %LOG%)
+python -m pip install -U pip wheel > "%LOG%" 2>&1
+if errorlevel 1 (
+  echo [ERREUR] Upgrade pip/wheel a echoue. Voir %LOG%.
+  goto VIEW_LOG
+)
+pip install -r requirements.txt >> "%LOG%" 2>&1
+if errorlevel 1 (
+  echo [ERREUR] Installation requirements a echoue. Voir %LOG%.
+  goto VIEW_LOG
+)
+
+echo.
+echo [2a] sentencepiece (necessaire pour CamemBERT/DrBERT)
+pip install --only-binary=:all: sentencepiece==0.1.99 >> "%LOG%" 2>&1
+if not errorlevel 1 set "SPM_MISSING=0"
+
+echo.
+echo [2b] Test imports (core)
+python -c "import pdfplumber,spacy,requests,transformers,torch,tokenizers,huggingface_hub,yaml,PyInstaller,sys,importlib.util as u; print('Core imports OK. sentencepiece=', bool(u.find_spec('sentencepiece')))"
+if errorlevel 1 (
+  echo [ERREUR] Echec imports Python de base. Voir %LOG%.
+  goto VIEW_LOG
+)
+
+echo.
+echo [3] Modele spaCy fr_core_news_lg
+if exist "%MODEL_DIR%\config.cfg" (
+  echo [OK] Modele local detecte: %MODEL_DIR%
+) else (
+  echo [INFO] Tentative A: python -m spacy download fr_core_news_lg
+  python -m spacy download fr_core_news_lg >> "%LOG%" 2>&1
+  if errorlevel 1 (
+    echo [INFO] Tentative B: pip install wheel officiel
+    pip install "%FR_WHEEL_URL%" >> "%LOG%" 2>&1
+    if errorlevel 1 (
+      echo [WARN] Echec installation du modele spaCy. Vous pourrez le telecharger via l'UI.
+    ) else (
+      echo [OK] Modele installe via wheel.
+    )
+  ) else (
+    echo [OK] Modele telecharge via spacy.
+  )
+)
+
+echo.
+echo [3bis] Pre-cache HuggingFace (accelere le 1er usage)
+if "%SPM_MISSING%"=="0" (
+  set "HF_CACHE=%LOCALAPPDATA%\Pseudonymiseur\models\hf_cache"
+  set "HF_HOME=%HF_CACHE%"
+  echo    Cache: %HF_CACHE%
+
+  set "HF_PRECACHE=%TEMP%\hf_precache.py"
+  > "%HF_PRECACHE%" echo import os
+  >>"%HF_PRECACHE%" echo os.environ['HF_HOME']=r'%HF_CACHE%'
+  >>"%HF_PRECACHE%" echo from transformers import AutoTokenizer, AutoModelForTokenClassification, AutoModel
+  >>"%HF_PRECACHE%" echo # Tokenizers
+  >>"%HF_PRECACHE%" echo AutoTokenizer.from_pretrained('Jean-Baptiste/camembert-ner')
+  >>"%HF_PRECACHE%" echo AutoTokenizer.from_pretrained('almanach/camembert-base-bio')
+  >>"%HF_PRECACHE%" echo AutoTokenizer.from_pretrained('Dr-BERT/DrBERT-7GB')
+  >>"%HF_PRECACHE%" echo # Models
+  >>"%HF_PRECACHE%" echo AutoModelForTokenClassification.from_pretrained('Jean-Baptiste/camembert-ner')
+  >>"%HF_PRECACHE%" echo AutoModel.from_pretrained('almanach/camembert-base-bio')
+  >>"%HF_PRECACHE%" echo AutoModel.from_pretrained('Dr-BERT/DrBERT-7GB')
+
+  python "%HF_PRECACHE%" >> "%LOG%" 2>&1
+  del /f /q "%HF_PRECACHE%" >nul 2>&1
+  if errorlevel 1 (echo [WARN] Pre-cache HF partiel. Voir %LOG%.) else (echo [OK] Pre-cache HF)
+) else (
+  echo [INFO] Pre-cache HF saute (sentencepiece manquant).
+)
+
+
+:MENU
+echo.
+echo ================== MENU ==================
+echo [A] Lancer l'application (UI)
+echo [B] Builder EXE onefile (sans console)
+echo [C] Builder EXE onedir (dev rapide)
+echo [X] Nettoyer (build/dist/spec/caches/logs)
+echo [V] Voir les 80 dernieres lignes du log
+echo [Q] Quitter (fenetre persiste)
+set /p CHOIX="Votre choix ? "
+if /I "%CHOIX%"=="A" goto RUN
+if /I "%CHOIX%"=="B" goto BUILD_ONEFILE
+if /I "%CHOIX%"=="C" goto BUILD_ONEDIR
+if /I "%CHOIX%"=="X" goto CLEAN_AGAIN
+if /I "%CHOIX%"=="V" goto VIEW_LOG
+if /I "%CHOIX%"=="Q" goto END
+echo Choix invalide.
+goto MENU
+
+:RUN
+echo.
+echo [RUN] Lancement de l'UI...
+python "%ENTRY%"
+echo.
+echo [INFO] L'UI s'est fermee. Retour menu.
+pause
+goto MENU
+
+:BUILD_ONEFILE
+echo.
+echo [BUILD] EXE onefile (sans console)
+taskkill /IM %EXENAME%.exe /F >nul 2>&1
+rmdir /s /q build dist out 2>nul
+set "PYI_COMMON=--clean --noconfirm --onefile --noconsole --name %EXENAME% --hidden-import=pdfplumber --hidden-import=pdfminer --hidden-import=pdfminer.six --hidden-import=cffi --hidden-import=_cffi_backend --hidden-import=cryptography --hidden-import=cryptography.hazmat.bindings._rust --hidden-import=sentencepiece --collect-binaries cryptography --collect-binaries cffi --collect-binaries sentencepiece --collect-data cryptography --collect-data pdfminer --collect-data pdfplumber --collect-data spacy --collect-all transformers --collect-all tokenizers --collect-all huggingface_hub --collect-data torch"
+set "PYI_MODEL="
+if exist "%MODEL_DIR%" set "PYI_MODEL=--add-data ""%MODEL_DIR%;%MODEL_DIR%"""
+echo [CMD] python -m PyInstaller %PYI_COMMON% %PYI_MODEL% "%ENTRY%"
+python -m PyInstaller %PYI_COMMON% %PYI_MODEL% "%ENTRY%" >> "%LOG%" 2>&1
+if errorlevel 1 (
+  echo [ERREUR] Build onefile. Voir %LOG% ci-dessous:
+  goto VIEW_LOG
+) else (
+  echo [OK] EXE : dist\%EXENAME%.exe
+  pause
+  goto MENU
+)
+
+:BUILD_ONEDIR
+echo.
+echo [BUILD] EXE onedir (dev rapide)
+set "PYI_MODEL="
+if exist "%MODEL_DIR%" set "PYI_MODEL=--add-data ""%MODEL_DIR%;%MODEL_DIR%"""
+python -m PyInstaller --clean --noconfirm --onedir --noconsole --name %EXENAME%_dev %PYI_MODEL% --hidden-import=pdfplumber --hidden-import=pdfminer --hidden-import=pdfminer.six --hidden-import=cffi --hidden-import=_cffi_backend --hidden-import=cryptography --hidden-import=cryptography.hazmat.bindings._rust --hidden-import=sentencepiece --collect-binaries cryptography --collect-binaries cffi --collect-binaries sentencepiece --collect-data cryptography --collect-data pdfminer --collect-data pdfplumber --collect-data spacy --collect-all transformers --collect-all tokenizers --collect-all huggingface_hub --collect-data torch "%ENTRY%" >> "%LOG%" 2>&1
+if errorlevel 1 (
+  echo [ERREUR] Build onedir. Voir %LOG% ci-dessous:
+  goto VIEW_LOG
+) else (
+  echo [OK] Dossier : dist\%EXENAME%_dev
+  pause
+  goto MENU
+)
+
+:CLEAN_AGAIN
+echo.
+echo [CLEAN] Suppression build/dist/out/*.spec/caches/logs
+if exist ".\build" rmdir /s /q ".\build" >nul 2>&1
+if exist ".\dist"  rmdir /s /q ".\dist"  >nul 2>&1
+if exist ".\out"   rmdir /s /q ".\out"   >nul 2>&1
+del /f /q *.spec build_log.txt 2>nul
+for /d %%D in (__pycache__ .pytest_cache .mypy_cache) do if exist "%%D" rmdir /s /q "%%D" 2>nul
+echo [CLEAN] OK
+pause
+goto MENU
+
+:VIEW_LOG
+echo.
+echo ===== Dernieres lignes de %LOG% =====
+if exist "%LOG%" (
+  powershell -NoLogo -NoProfile -Command "Get-Content -Path '%LOG%' -Tail 80"
+) else (
+  echo (pas de log pour l'instant)
+)
+echo =====================================
+pause
+goto MENU
+
+:END
+echo.
+echo Fin du script. La fenetre reste ouverte (mode persistant).