cjber commited on
Commit
8f48ef8
·
1 Parent(s): 8e1b43a

add pii removal

Browse files
Files changed (1) hide show
  1. planning_ai/nodes/map_node.py +16 -0
planning_ai/nodes/map_node.py CHANGED
@@ -1,10 +1,26 @@
1
  from langgraph.constants import Send
 
 
2
 
3
  from planning_ai.chains.map_chain import map_chain
4
  from planning_ai.states import DocumentState, OverallState
5
 
 
 
 
 
 
 
 
 
 
 
 
 
 
6
 
7
  def generate_summary(state: DocumentState):
 
8
  response = map_chain.invoke({"context": state["document"]})
9
  return {
10
  "summaries": [
 
1
  from langgraph.constants import Send
2
+ from presidio_analyzer import AnalyzerEngine
3
+ from presidio_anonymizer import AnonymizerEngine
4
 
5
  from planning_ai.chains.map_chain import map_chain
6
  from planning_ai.states import DocumentState, OverallState
7
 
8
+ anonymizer = AnonymizerEngine()
9
+ analyzer = AnalyzerEngine()
10
+
11
+
12
+ def remove_pii(document: str):
13
+ results = analyzer.analyze(
14
+ text=document,
15
+ entities=["PERSON", "PHONE_NUMBER", "EMAIL_ADDRESS"],
16
+ language="en",
17
+ )
18
+ document = anonymizer.anonymize(text=document, analyzer_results=results)
19
+ return document
20
+
21
 
22
  def generate_summary(state: DocumentState):
23
+ state["document"] = remove_pii(state["document"])
24
  response = map_chain.invoke({"context": state["document"]})
25
  return {
26
  "summaries": [