karthikmulugu08 commited on
Commit
e6ed707
·
verified ·
1 Parent(s): 72a5cdd

Fix: strip non-ASCII chars before sending to llama-server

Browse files
Files changed (1) hide show
  1. app.py +7 -1
app.py CHANGED
@@ -78,6 +78,12 @@ def _read_file(path: str) -> str:
78
  return p.read_text(encoding="utf-8", errors="ignore")
79
 
80
 
 
 
 
 
 
 
81
  def _build_full_context() -> str:
82
  """
83
  Build safety-check context via targeted retrieval for medications and allergies.
@@ -105,7 +111,7 @@ def _build_full_context() -> str:
105
  seen.add(uid)
106
  ordered.append(c)
107
 
108
- parts = [f"[{c.source}]\n{c.text}" for c in ordered]
109
  return "\n\n---\n\n".join(parts)
110
 
111
 
 
78
  return p.read_text(encoding="utf-8", errors="ignore")
79
 
80
 
81
+ def _sanitize(text: str) -> str:
82
+ """Replace non-ASCII chars that break JSON serialization in llama-server."""
83
+ import re
84
+ return re.sub(r'[^\x00-\x7F]', '-', text)
85
+
86
+
87
  def _build_full_context() -> str:
88
  """
89
  Build safety-check context via targeted retrieval for medications and allergies.
 
111
  seen.add(uid)
112
  ordered.append(c)
113
 
114
+ parts = [f"[{c.source}]\n{_sanitize(c.text)}" for c in ordered]
115
  return "\n\n---\n\n".join(parts)
116
 
117