Spaces:
Sleeping
Sleeping
Upload src\intent.py with huggingface_hub
Browse files- src//intent.py +52 -0
src//intent.py
ADDED
|
@@ -0,0 +1,52 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
"""
|
| 2 |
+
Intent inference for notebook cells
|
| 3 |
+
"""
|
| 4 |
+
|
| 5 |
+
from typing import List, Optional
|
| 6 |
+
import re
|
| 7 |
+
from src.models import ContextUnit, CellType
|
| 8 |
+
|
| 9 |
+
|
| 10 |
+
class ContextThreadEnricher:
|
| 11 |
+
"""Enriches context threads with intent information."""
|
| 12 |
+
|
| 13 |
+
def __init__(self, infer_intents: bool = True):
|
| 14 |
+
self.infer_intents = infer_intents
|
| 15 |
+
|
| 16 |
+
def enrich(self, thread):
|
| 17 |
+
"""Enrich the thread with intents."""
|
| 18 |
+
if not self.infer_intents:
|
| 19 |
+
return thread
|
| 20 |
+
|
| 21 |
+
for unit in thread.units:
|
| 22 |
+
if unit.intent == "[Pending intent inference]":
|
| 23 |
+
unit.intent = self._infer_intent_heuristic(unit)
|
| 24 |
+
|
| 25 |
+
return thread
|
| 26 |
+
|
| 27 |
+
def _infer_intent_heuristic(self, unit: ContextUnit) -> str:
|
| 28 |
+
"""Infer intent using heuristics."""
|
| 29 |
+
source = unit.cell.source.lower()
|
| 30 |
+
|
| 31 |
+
# Data loading
|
| 32 |
+
if any(keyword in source for keyword in ['read_csv', 'read_excel', 'load', 'open']):
|
| 33 |
+
return "Load data from file"
|
| 34 |
+
|
| 35 |
+
# Data cleaning
|
| 36 |
+
if any(keyword in source for keyword in ['dropna', 'fillna', 'clean', 'remove', 'filter']):
|
| 37 |
+
return "Clean and preprocess data"
|
| 38 |
+
|
| 39 |
+
# Analysis/Modeling
|
| 40 |
+
if any(keyword in source for keyword in ['fit', 'predict', 'train', 'model', 'regression']):
|
| 41 |
+
return "Build and train model"
|
| 42 |
+
|
| 43 |
+
# Visualization
|
| 44 |
+
if any(keyword in source for keyword in ['plot', 'chart', 'graph', 'visualize', 'show']):
|
| 45 |
+
return "Create visualization"
|
| 46 |
+
|
| 47 |
+
# Statistics
|
| 48 |
+
if any(keyword in source for keyword in ['mean', 'std', 'sum', 'count', 'describe']):
|
| 49 |
+
return "Compute statistics"
|
| 50 |
+
|
| 51 |
+
# Default
|
| 52 |
+
return "Execute code"
|