Spaces:
Sleeping
Sleeping
Update src/cv_parsing_agents.py
Browse files- src/cv_parsing_agents.py +115 -63
src/cv_parsing_agents.py
CHANGED
|
@@ -7,23 +7,8 @@ import logging
|
|
| 7 |
|
| 8 |
logger = logging.getLogger(__name__)
|
| 9 |
|
| 10 |
-
|
| 11 |
-
|
| 12 |
-
CREW_POOL_AVAILABLE = True
|
| 13 |
-
logger.info("✅ crew_pool importé avec succès")
|
| 14 |
-
except ImportError as e:
|
| 15 |
-
logger.error(f"❌ Erreur import crew_pool: {e}")
|
| 16 |
-
CREW_POOL_AVAILABLE = False
|
| 17 |
-
analyse_cv = None
|
| 18 |
-
|
| 19 |
-
try:
|
| 20 |
-
from src.config import load_pdf
|
| 21 |
-
CONFIG_AVAILABLE = True
|
| 22 |
-
logger.info("✅ config importé avec succès")
|
| 23 |
-
except ImportError as e:
|
| 24 |
-
logger.error(f"❌ Erreur import config: {e}")
|
| 25 |
-
CONFIG_AVAILABLE = False
|
| 26 |
-
load_pdf = None
|
| 27 |
|
| 28 |
def clean_dict_keys(data):
|
| 29 |
"""
|
|
@@ -51,64 +36,76 @@ class OptimizedCvParserAgent:
|
|
| 51 |
"""
|
| 52 |
|
| 53 |
def __init__(self, pdf_path: str):
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 54 |
if not pdf_path or not isinstance(pdf_path, str):
|
| 55 |
raise ValueError("Le chemin du fichier PDF doit être une chaîne non vide")
|
| 56 |
|
| 57 |
self.pdf_path = pdf_path
|
| 58 |
-
|
| 59 |
-
if not CREW_POOL_AVAILABLE:
|
| 60 |
-
logger.warning("CrewAI crew_pool non disponible - mode dégradé")
|
| 61 |
-
if not CONFIG_AVAILABLE:
|
| 62 |
-
logger.warning("Module config non disponible - mode dégradé")
|
| 63 |
|
| 64 |
def process(self) -> dict:
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 65 |
logger.info(f"Début du traitement optimisé du CV : {self.pdf_path}")
|
| 66 |
|
| 67 |
if not os.path.exists(self.pdf_path):
|
| 68 |
-
|
| 69 |
-
|
| 70 |
-
|
| 71 |
-
if not
|
| 72 |
-
|
| 73 |
-
|
| 74 |
-
|
| 75 |
-
|
| 76 |
-
|
| 77 |
-
if not cv_text_content or not cv_text_content.strip():
|
| 78 |
-
logger.error("Le PDF semble vide ou illisible")
|
| 79 |
-
return self._create_fallback_data()
|
| 80 |
-
|
| 81 |
-
logger.info(f"PDF chargé, {len(cv_text_content)} caractères extraits")
|
| 82 |
-
|
| 83 |
-
crew_output = analyse_cv(cv_text_content)
|
| 84 |
-
|
| 85 |
-
if not crew_output or not hasattr(crew_output, 'raw') or not crew_output.raw.strip():
|
| 86 |
-
logger.error("L'analyse par le crew n'a pas retourné de résultat.")
|
| 87 |
-
return self._create_fallback_data()
|
| 88 |
-
raw_string = crew_output.raw
|
| 89 |
-
logger.info(f"Résultat brut du crew optimisé: {raw_string[:200]}...")
|
| 90 |
-
json_string_cleaned = self._clean_json_string(raw_string)
|
| 91 |
-
profile_data = json.loads(json_string_cleaned)
|
| 92 |
-
logger.info("Parsing JSON optimisé réussi")
|
| 93 |
-
|
| 94 |
-
optimized_data = self._validate_and_enhance_data(profile_data)
|
| 95 |
-
|
| 96 |
-
return clean_dict_keys(optimized_data)
|
| 97 |
|
| 98 |
-
|
| 99 |
-
|
| 100 |
-
if 'crew_output' in locals():
|
| 101 |
-
logger.error(f"Données brutes reçues : {crew_output.raw}")
|
| 102 |
-
return self._create_fallback_data()
|
| 103 |
|
| 104 |
-
|
| 105 |
-
|
| 106 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 107 |
|
| 108 |
def _validate_and_enhance_data(self, profile_data: dict) -> dict:
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 109 |
if not isinstance(profile_data, dict) or "candidat" not in profile_data:
|
| 110 |
-
|
| 111 |
-
return self._create_fallback_data()
|
| 112 |
|
| 113 |
candidat = profile_data["candidat"]
|
| 114 |
|
|
@@ -116,12 +113,15 @@ class OptimizedCvParserAgent:
|
|
| 116 |
"informations_personnelles", "compétences", "expériences",
|
| 117 |
"projets", "formations", "reconversion"
|
| 118 |
]
|
|
|
|
| 119 |
for section in required_sections:
|
| 120 |
if section not in candidat or not candidat[section]:
|
| 121 |
logger.warning(f"Section manquante ou vide: {section}")
|
| 122 |
candidat[section] = self._get_default_section_data(section)
|
|
|
|
| 123 |
self._normalize_competences(candidat.get("compétences", {}))
|
| 124 |
self._normalize_experiences(candidat.get("expériences", []))
|
|
|
|
| 125 |
logger.info("Validation et enrichissement des données terminés")
|
| 126 |
return profile_data
|
| 127 |
|
|
@@ -129,10 +129,12 @@ class OptimizedCvParserAgent:
|
|
| 129 |
"""Normalise la section compétences"""
|
| 130 |
if not isinstance(competences, dict):
|
| 131 |
return
|
|
|
|
| 132 |
if "hard_skills" not in competences:
|
| 133 |
competences["hard_skills"] = []
|
| 134 |
if "soft_skills" not in competences:
|
| 135 |
competences["soft_skills"] = []
|
|
|
|
| 136 |
competences["hard_skills"] = [skill.strip() for skill in competences["hard_skills"] if skill and skill.strip()]
|
| 137 |
competences["soft_skills"] = [skill.strip() for skill in competences["soft_skills"] if skill and skill.strip()]
|
| 138 |
|
|
@@ -140,10 +142,13 @@ class OptimizedCvParserAgent:
|
|
| 140 |
"""Normalise la section expériences"""
|
| 141 |
if not isinstance(experiences, list):
|
| 142 |
return
|
|
|
|
| 143 |
required_fields = ["Poste", "Entreprise", "start_date", "end_date", "responsabilités"]
|
|
|
|
| 144 |
for exp in experiences:
|
| 145 |
if not isinstance(exp, dict):
|
| 146 |
continue
|
|
|
|
| 147 |
for field in required_fields:
|
| 148 |
if field not in exp or exp[field] in [None, "", []]:
|
| 149 |
exp[field] = "Non spécifié" if field != "responsabilités" else []
|
|
@@ -175,6 +180,41 @@ class OptimizedCvParserAgent:
|
|
| 175 |
return defaults.get(section, {})
|
| 176 |
|
| 177 |
def _clean_json_string(self, raw_string: str) -> str:
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 178 |
json_string_cleaned = raw_string.strip()
|
| 179 |
|
| 180 |
if '```' in raw_string:
|
|
@@ -192,6 +232,12 @@ class OptimizedCvParserAgent:
|
|
| 192 |
return json_string_cleaned
|
| 193 |
|
| 194 |
def get_processing_stats(self) -> dict:
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 195 |
return {
|
| 196 |
"optimization_enabled": True,
|
| 197 |
"section_based_processing": True,
|
|
@@ -208,5 +254,11 @@ class CvParserAgent(OptimizedCvParserAgent):
|
|
| 208 |
|
| 209 |
if __name__ == "__main__":
|
| 210 |
logger.info("Test du module cv_parsing_agents optimisé")
|
| 211 |
-
|
| 212 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 7 |
|
| 8 |
logger = logging.getLogger(__name__)
|
| 9 |
|
| 10 |
+
from src.crew.crew_pool import analyse_cv
|
| 11 |
+
from src.config import load_pdf
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 12 |
|
| 13 |
def clean_dict_keys(data):
|
| 14 |
"""
|
|
|
|
| 36 |
"""
|
| 37 |
|
| 38 |
def __init__(self, pdf_path: str):
|
| 39 |
+
"""
|
| 40 |
+
Initialise l'agent de parsing de CV optimisé.
|
| 41 |
+
|
| 42 |
+
Args:
|
| 43 |
+
pdf_path (str): Chemin vers le fichier PDF à traiter
|
| 44 |
+
|
| 45 |
+
Raises:
|
| 46 |
+
ValueError: Si le chemin du fichier est invalide
|
| 47 |
+
"""
|
| 48 |
if not pdf_path or not isinstance(pdf_path, str):
|
| 49 |
raise ValueError("Le chemin du fichier PDF doit être une chaîne non vide")
|
| 50 |
|
| 51 |
self.pdf_path = pdf_path
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 52 |
|
| 53 |
def process(self) -> dict:
|
| 54 |
+
"""
|
| 55 |
+
Traite le fichier PDF pour en extraire le contenu sous forme de JSON optimisé.
|
| 56 |
+
|
| 57 |
+
Returns:
|
| 58 |
+
dict: Dictionnaire contenant les données extraites du CV
|
| 59 |
+
|
| 60 |
+
Raises:
|
| 61 |
+
FileNotFoundError: Si le fichier PDF n'existe pas
|
| 62 |
+
ValueError: Si le PDF est vide ou illisible
|
| 63 |
+
json.JSONDecodeError: Si le résultat n'est pas un JSON valide
|
| 64 |
+
Exception: Pour toute autre erreur de traitement
|
| 65 |
+
"""
|
| 66 |
logger.info(f"Début du traitement optimisé du CV : {self.pdf_path}")
|
| 67 |
|
| 68 |
if not os.path.exists(self.pdf_path):
|
| 69 |
+
raise FileNotFoundError(f"Fichier PDF non trouvé: {self.pdf_path}")
|
| 70 |
+
|
| 71 |
+
cv_text_content = load_pdf(self.pdf_path)
|
| 72 |
+
if not cv_text_content or not cv_text_content.strip():
|
| 73 |
+
raise ValueError("Le PDF semble vide ou illisible")
|
| 74 |
+
|
| 75 |
+
logger.info(f"PDF chargé, {len(cv_text_content)} caractères extraits")
|
| 76 |
+
|
| 77 |
+
crew_output = analyse_cv(cv_text_content)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 78 |
|
| 79 |
+
if not crew_output or not hasattr(crew_output, 'raw') or not crew_output.raw.strip():
|
| 80 |
+
raise Exception("L'analyse par le crew n'a pas retourné de résultat.")
|
|
|
|
|
|
|
|
|
|
| 81 |
|
| 82 |
+
raw_string = crew_output.raw
|
| 83 |
+
logger.info(f"Résultat brut du crew optimisé: {raw_string[:200]}...")
|
| 84 |
+
|
| 85 |
+
json_string_cleaned = self._clean_json_string(raw_string)
|
| 86 |
+
|
| 87 |
+
profile_data = json.loads(json_string_cleaned)
|
| 88 |
+
logger.info("Parsing JSON optimisé réussi")
|
| 89 |
+
|
| 90 |
+
optimized_data = self._validate_and_enhance_data(profile_data)
|
| 91 |
+
|
| 92 |
+
return clean_dict_keys(optimized_data)
|
| 93 |
|
| 94 |
def _validate_and_enhance_data(self, profile_data: dict) -> dict:
|
| 95 |
+
"""
|
| 96 |
+
Valide et enrichit les données extraites du CV.
|
| 97 |
+
|
| 98 |
+
Args:
|
| 99 |
+
profile_data (dict): Données brutes extraites
|
| 100 |
+
|
| 101 |
+
Returns:
|
| 102 |
+
dict: Données validées et enrichies
|
| 103 |
+
|
| 104 |
+
Raises:
|
| 105 |
+
ValueError: Si la structure de données est invalide
|
| 106 |
+
"""
|
| 107 |
if not isinstance(profile_data, dict) or "candidat" not in profile_data:
|
| 108 |
+
raise ValueError("Structure de données invalide - clé 'candidat' manquante")
|
|
|
|
| 109 |
|
| 110 |
candidat = profile_data["candidat"]
|
| 111 |
|
|
|
|
| 113 |
"informations_personnelles", "compétences", "expériences",
|
| 114 |
"projets", "formations", "reconversion"
|
| 115 |
]
|
| 116 |
+
|
| 117 |
for section in required_sections:
|
| 118 |
if section not in candidat or not candidat[section]:
|
| 119 |
logger.warning(f"Section manquante ou vide: {section}")
|
| 120 |
candidat[section] = self._get_default_section_data(section)
|
| 121 |
+
|
| 122 |
self._normalize_competences(candidat.get("compétences", {}))
|
| 123 |
self._normalize_experiences(candidat.get("expériences", []))
|
| 124 |
+
|
| 125 |
logger.info("Validation et enrichissement des données terminés")
|
| 126 |
return profile_data
|
| 127 |
|
|
|
|
| 129 |
"""Normalise la section compétences"""
|
| 130 |
if not isinstance(competences, dict):
|
| 131 |
return
|
| 132 |
+
|
| 133 |
if "hard_skills" not in competences:
|
| 134 |
competences["hard_skills"] = []
|
| 135 |
if "soft_skills" not in competences:
|
| 136 |
competences["soft_skills"] = []
|
| 137 |
+
|
| 138 |
competences["hard_skills"] = [skill.strip() for skill in competences["hard_skills"] if skill and skill.strip()]
|
| 139 |
competences["soft_skills"] = [skill.strip() for skill in competences["soft_skills"] if skill and skill.strip()]
|
| 140 |
|
|
|
|
| 142 |
"""Normalise la section expériences"""
|
| 143 |
if not isinstance(experiences, list):
|
| 144 |
return
|
| 145 |
+
|
| 146 |
required_fields = ["Poste", "Entreprise", "start_date", "end_date", "responsabilités"]
|
| 147 |
+
|
| 148 |
for exp in experiences:
|
| 149 |
if not isinstance(exp, dict):
|
| 150 |
continue
|
| 151 |
+
|
| 152 |
for field in required_fields:
|
| 153 |
if field not in exp or exp[field] in [None, "", []]:
|
| 154 |
exp[field] = "Non spécifié" if field != "responsabilités" else []
|
|
|
|
| 180 |
return defaults.get(section, {})
|
| 181 |
|
| 182 |
def _clean_json_string(self, raw_string: str) -> str:
|
| 183 |
+
"""
|
| 184 |
+
Nettoie une chaîne JSON brute en supprimant les blocs de code markdown.
|
| 185 |
+
|
| 186 |
+
Args:
|
| 187 |
+
raw_string (str): Chaîne brute à nettoyer
|
| 188 |
+
|
| 189 |
+
Returns:
|
| 190 |
+
str: Chaîne JSON nettoyée
|
| 191 |
+
"""
|
| 192 |
+
json_string_cleaned = raw_string.strip()
|
| 193 |
+
|
| 194 |
+
if '```' in raw_string:
|
| 195 |
+
try:
|
| 196 |
+
if '```json' in raw_string:
|
| 197 |
+
json_part = raw_string.split('```json')[1].split('```')[0]
|
| 198 |
+
json_string_cleaned = json_part.strip()
|
| 199 |
+
else:
|
| 200 |
+
parts = raw_string.split('```')
|
| 201 |
+
if len(parts) >= 3:
|
| 202 |
+
json_string_cleaned = parts[1].strip()
|
| 203 |
+
except IndexError:
|
| 204 |
+
logger.warning("Format de code block détecté mais mal formé")
|
| 205 |
+
|
| 206 |
+
return json_string_cleaned
|
| 207 |
+
|
| 208 |
+
def _clean_json_string(self, raw_string: str) -> str:
|
| 209 |
+
"""
|
| 210 |
+
Nettoie une chaîne JSON brute en supprimant les blocs de code markdown.
|
| 211 |
+
|
| 212 |
+
Args:
|
| 213 |
+
raw_string (str): Chaîne brute à nettoyer
|
| 214 |
+
|
| 215 |
+
Returns:
|
| 216 |
+
str: Chaîne JSON nettoyée
|
| 217 |
+
"""
|
| 218 |
json_string_cleaned = raw_string.strip()
|
| 219 |
|
| 220 |
if '```' in raw_string:
|
|
|
|
| 232 |
return json_string_cleaned
|
| 233 |
|
| 234 |
def get_processing_stats(self) -> dict:
|
| 235 |
+
"""
|
| 236 |
+
Retourne des statistiques sur l'optimisation du traitement.
|
| 237 |
+
|
| 238 |
+
Returns:
|
| 239 |
+
dict: Statistiques d'optimisation
|
| 240 |
+
"""
|
| 241 |
return {
|
| 242 |
"optimization_enabled": True,
|
| 243 |
"section_based_processing": True,
|
|
|
|
| 254 |
|
| 255 |
if __name__ == "__main__":
|
| 256 |
logger.info("Test du module cv_parsing_agents optimisé")
|
| 257 |
+
|
| 258 |
+
try:
|
| 259 |
+
agent = OptimizedCvParserAgent("/tmp/test.pdf")
|
| 260 |
+
stats = agent.get_processing_stats()
|
| 261 |
+
logger.info("✅ OptimizedCvParserAgent créé avec succès")
|
| 262 |
+
logger.info(f"✅ Statistiques d'optimisation: {stats}")
|
| 263 |
+
except Exception as e:
|
| 264 |
+
logger.error(f"❌ Erreur création OptimizedCvParserAgent: {e}")
|