Spaces:
Runtime error
Runtime error
Update conversation_storyline/schemas.py
Browse files
conversation_storyline/schemas.py
CHANGED
|
@@ -1,31 +1,92 @@
|
|
| 1 |
-
|
| 2 |
-
from
|
| 3 |
-
|
| 4 |
-
|
| 5 |
-
|
| 6 |
-
|
| 7 |
-
|
| 8 |
-
|
| 9 |
-
|
| 10 |
-
|
| 11 |
-
|
| 12 |
-
|
| 13 |
-
|
| 14 |
-
|
| 15 |
-
|
| 16 |
-
|
| 17 |
-
|
| 18 |
-
|
| 19 |
-
|
| 20 |
-
|
| 21 |
-
|
| 22 |
-
|
| 23 |
-
|
| 24 |
-
|
| 25 |
-
|
| 26 |
-
|
| 27 |
-
|
| 28 |
-
|
| 29 |
-
|
| 30 |
-
|
| 31 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
# conversation_storyline/schemas.py
|
| 2 |
+
from __future__ import annotations
|
| 3 |
+
|
| 4 |
+
from typing import List, Optional, Literal, Dict, Any
|
| 5 |
+
from pydantic import BaseModel, Field, ConfigDict
|
| 6 |
+
|
| 7 |
+
InteractionType = Literal[
|
| 8 |
+
"statement",
|
| 9 |
+
"question",
|
| 10 |
+
"answer",
|
| 11 |
+
"interruption",
|
| 12 |
+
"agreement",
|
| 13 |
+
"disagreement",
|
| 14 |
+
]
|
| 15 |
+
|
| 16 |
+
class RawMessage(BaseModel):
|
| 17 |
+
"""
|
| 18 |
+
Representa un mensaje crudo tras la ingesta (TXT/CSV o texto pegado).
|
| 19 |
+
"""
|
| 20 |
+
model_config = ConfigDict(extra="ignore")
|
| 21 |
+
|
| 22 |
+
id: int = Field(..., description="ID secuencial global (0..n-1).")
|
| 23 |
+
speaker: str = Field(..., description="Nombre normalizado del hablante.")
|
| 24 |
+
content: str = Field(..., description="Texto original del turno.")
|
| 25 |
+
timestamp: Optional[str] = Field(None, description="Timestamp opcional si existe.")
|
| 26 |
+
|
| 27 |
+
|
| 28 |
+
class InteractionNode(BaseModel):
|
| 29 |
+
"""
|
| 30 |
+
Turno estructurado (salida del m贸dulo SIE / LLM).
|
| 31 |
+
"""
|
| 32 |
+
model_config = ConfigDict(extra="ignore")
|
| 33 |
+
|
| 34 |
+
id: int = Field(..., description="ID secuencial global del turno.")
|
| 35 |
+
speaker: str = Field(..., description="Nombre normalizado del hablante.")
|
| 36 |
+
content_summary: str = Field(..., description="Resumen breve (1 frase).")
|
| 37 |
+
|
| 38 |
+
reply_to_id: Optional[int] = Field(
|
| 39 |
+
None,
|
| 40 |
+
description="ID del mensaje al que responde. Null si inicia hilo o no se infiere con confianza.",
|
| 41 |
+
)
|
| 42 |
+
|
| 43 |
+
topic_label: str = Field(..., description="Etiqueta breve del tema (ej: 'Presupuesto').")
|
| 44 |
+
is_topic_shift: bool = Field(False, description="True si hay cambio sustancial de tema.")
|
| 45 |
+
sentiment_score: float = Field(..., ge=-1.0, le=1.0, description="Sentimiento [-1, 1].")
|
| 46 |
+
interaction_type: InteractionType = Field(..., description="Tipo de acto de habla.")
|
| 47 |
+
|
| 48 |
+
confidence: float = Field(0.75, ge=0.0, le=1.0, description="Confianza global del modelo.")
|
| 49 |
+
target_speaker: Optional[str] = Field(None, description="Hablante objetivo si reply_to_id se resuelve.")
|
| 50 |
+
|
| 51 |
+
|
| 52 |
+
class ConversationSegment(BaseModel):
|
| 53 |
+
"""
|
| 54 |
+
Salida estructurada por ventana/segmento de an谩lisis.
|
| 55 |
+
"""
|
| 56 |
+
model_config = ConfigDict(extra="ignore")
|
| 57 |
+
|
| 58 |
+
interactions: List[InteractionNode] = Field(..., description="Interacciones estructuradas en este segmento.")
|
| 59 |
+
active_participants: List[str] = Field(default_factory=list, description="Participantes activos del segmento.")
|
| 60 |
+
dominant_topic: str = Field(..., description="Tema dominante del segmento.")
|
| 61 |
+
notes: Optional[str] = Field(None, description="Notas opcionales.")
|
| 62 |
+
|
| 63 |
+
|
| 64 |
+
class GraphExport(BaseModel):
|
| 65 |
+
"""
|
| 66 |
+
Exportaci贸n tipo D3: nodes/links.
|
| 67 |
+
"""
|
| 68 |
+
model_config = ConfigDict(extra="ignore")
|
| 69 |
+
|
| 70 |
+
nodes: List[Dict[str, Any]]
|
| 71 |
+
links: List[Dict[str, Any]]
|
| 72 |
+
|
| 73 |
+
|
| 74 |
+
class ReplyToPick(BaseModel):
|
| 75 |
+
"""
|
| 76 |
+
Output estructurado para elegir reply_to_id entre candidatos (Top-K).
|
| 77 |
+
"""
|
| 78 |
+
model_config = ConfigDict(extra="ignore")
|
| 79 |
+
|
| 80 |
+
reply_to_id: Optional[int] = Field(None, description="ID elegido (o null).")
|
| 81 |
+
confidence: float = Field(0.75, ge=0.0, le=1.0, description="Confianza del pick.")
|
| 82 |
+
rationale: Optional[str] = Field(None, description="Breve justificaci贸n.")
|
| 83 |
+
|
| 84 |
+
|
| 85 |
+
class TopicLabeling(BaseModel):
|
| 86 |
+
"""
|
| 87 |
+
Output estructurado para etiquetar un segmento tem谩tico detectado (change-points).
|
| 88 |
+
"""
|
| 89 |
+
model_config = ConfigDict(extra="ignore")
|
| 90 |
+
|
| 91 |
+
topic_label: str = Field(..., description="Etiqueta breve del tema (3-6 palabras).")
|
| 92 |
+
keywords: List[str] = Field(default_factory=list, description="Keywords representativas.")
|