Update app.py
Browse files
app.py
CHANGED
|
@@ -1,50 +1,12 @@
|
|
| 1 |
import streamlit as st
|
| 2 |
import json
|
| 3 |
|
| 4 |
-
|
| 5 |
-
COMMON_NAME_KEYS = ["user", "username", "name", "fullName", "firstName", "lastName"]
|
| 6 |
LOGIN_KEYS = ["lastLogin", "login", "loggedIn", "lastLoggedIn", "last_login", "last_logged_in"]
|
| 7 |
|
| 8 |
-
# --- Session state setup ---
|
| 9 |
-
if "json_data" not in st.session_state:
|
| 10 |
-
st.session_state.json_data = {}
|
| 11 |
-
if "messages" not in st.session_state:
|
| 12 |
-
st.session_state.messages = []
|
| 13 |
-
if "temp_input" not in st.session_state:
|
| 14 |
-
st.session_state.temp_input = ""
|
| 15 |
-
if "files_loaded" not in st.session_state:
|
| 16 |
-
st.session_state.files_loaded = False
|
| 17 |
-
|
| 18 |
-
st.set_page_config(page_title="Instant JSON Q&A", layout="wide")
|
| 19 |
-
st.title("Instant JSON-Backed AI Q&A (No More Clarifying Loops)")
|
| 20 |
-
|
| 21 |
-
# --- Upload area ---
|
| 22 |
-
st.sidebar.header("Upload Multiple JSON Files")
|
| 23 |
-
uploaded_files = st.sidebar.file_uploader(
|
| 24 |
-
"Choose one or more JSON files", type="json", accept_multiple_files=True
|
| 25 |
-
)
|
| 26 |
-
|
| 27 |
-
if uploaded_files and not st.session_state.files_loaded:
|
| 28 |
-
st.session_state.json_data.clear()
|
| 29 |
-
for f in uploaded_files:
|
| 30 |
-
try:
|
| 31 |
-
content = json.load(f)
|
| 32 |
-
st.session_state.json_data[f.name] = content
|
| 33 |
-
st.sidebar.success(f"Loaded: {f.name}")
|
| 34 |
-
except Exception as e:
|
| 35 |
-
st.sidebar.error(f"Error reading {f.name}: {e}")
|
| 36 |
-
st.session_state.messages = []
|
| 37 |
-
st.session_state.files_loaded = True
|
| 38 |
-
elif not uploaded_files:
|
| 39 |
-
st.session_state.json_data.clear()
|
| 40 |
-
st.session_state.files_loaded = False
|
| 41 |
-
|
| 42 |
-
# --- Robust normalization function for matching ---
|
| 43 |
def normalize(s):
|
| 44 |
-
# Lowercase, remove underscores/dashes, strip, collapse double spaces
|
| 45 |
return str(s).lower().replace("_", " ").replace("-", " ").replace(".", " ").strip()
|
| 46 |
|
| 47 |
-
# --- Recursive user search with robust matching ---
|
| 48 |
def recursive_find_user(target_name):
|
| 49 |
matches = []
|
| 50 |
norm_target = normalize(target_name)
|
|
@@ -52,7 +14,8 @@ def recursive_find_user(target_name):
|
|
| 52 |
def _search(obj, path):
|
| 53 |
if isinstance(obj, dict):
|
| 54 |
for k, v in obj.items():
|
| 55 |
-
|
|
|
|
| 56 |
login_info = {}
|
| 57 |
for lk in LOGIN_KEYS:
|
| 58 |
if lk in obj:
|
|
@@ -64,6 +27,21 @@ def recursive_find_user(target_name):
|
|
| 64 |
"file": file_name,
|
| 65 |
"login_info": login_info
|
| 66 |
})
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 67 |
_search(v, path + [k])
|
| 68 |
elif isinstance(obj, list):
|
| 69 |
for idx, item in enumerate(obj):
|
|
@@ -71,7 +49,6 @@ def recursive_find_user(target_name):
|
|
| 71 |
_search(data, [])
|
| 72 |
return matches
|
| 73 |
|
| 74 |
-
# --- Show all user keys/values (for debug) ---
|
| 75 |
def show_all_users():
|
| 76 |
found = []
|
| 77 |
for file_name, data in st.session_state.json_data.items():
|
|
@@ -79,7 +56,12 @@ def show_all_users():
|
|
| 79 |
if isinstance(obj, dict):
|
| 80 |
for k, v in obj.items():
|
| 81 |
if k in COMMON_NAME_KEYS:
|
| 82 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 83 |
recursive(v, path + [k])
|
| 84 |
elif isinstance(obj, list):
|
| 85 |
for idx, item in enumerate(obj):
|
|
@@ -87,10 +69,8 @@ def show_all_users():
|
|
| 87 |
recursive(data, [])
|
| 88 |
return found
|
| 89 |
|
| 90 |
-
# --- Main Q&A logic ---
|
| 91 |
import re
|
| 92 |
def handle_user_query(query):
|
| 93 |
-
# Try to extract a likely user name from the query
|
| 94 |
patterns = [
|
| 95 |
r"(?:last\s*login.*?for|when\s+did)\s+([a-zA-Z0-9 _\-\.]+)",
|
| 96 |
r"when\s+was\s+([a-zA-Z0-9 _\-\.]+)\s+last\s+(?:login|logged\s*in)",
|
|
@@ -104,7 +84,6 @@ def handle_user_query(query):
|
|
| 104 |
found_name = m.group(1).strip()
|
| 105 |
break
|
| 106 |
if not found_name:
|
| 107 |
-
# Fallback: look for first capitalized word or phrase
|
| 108 |
m = re.search(r"([A-Z][a-z]+(?: [A-Z][a-z]+)*)", query)
|
| 109 |
if m:
|
| 110 |
found_name = m.group(1).strip()
|
|
@@ -122,7 +101,37 @@ def handle_user_query(query):
|
|
| 122 |
else:
|
| 123 |
return "Sorry, I can only answer direct user info queries (e.g., 'When did Bob the Builder last login?')."
|
| 124 |
|
| 125 |
-
# ---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 126 |
st.markdown("### Ask about any user directly (e.g. 'When did Bob the Builder last login?')")
|
| 127 |
for msg in st.session_state.messages:
|
| 128 |
if msg["role"] == "user":
|
|
|
|
| 1 |
import streamlit as st
|
| 2 |
import json
|
| 3 |
|
| 4 |
+
COMMON_NAME_KEYS = ["user", "username", "name", "fullName", "firstName", "lastName", "customer"]
|
|
|
|
| 5 |
LOGIN_KEYS = ["lastLogin", "login", "loggedIn", "lastLoggedIn", "last_login", "last_logged_in"]
|
| 6 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 7 |
def normalize(s):
|
|
|
|
| 8 |
return str(s).lower().replace("_", " ").replace("-", " ").replace(".", " ").strip()
|
| 9 |
|
|
|
|
| 10 |
def recursive_find_user(target_name):
|
| 11 |
matches = []
|
| 12 |
norm_target = normalize(target_name)
|
|
|
|
| 14 |
def _search(obj, path):
|
| 15 |
if isinstance(obj, dict):
|
| 16 |
for k, v in obj.items():
|
| 17 |
+
# Direct match for string value
|
| 18 |
+
if k in COMMON_NAME_KEYS and isinstance(v, str) and norm_target in normalize(v):
|
| 19 |
login_info = {}
|
| 20 |
for lk in LOGIN_KEYS:
|
| 21 |
if lk in obj:
|
|
|
|
| 27 |
"file": file_name,
|
| 28 |
"login_info": login_info
|
| 29 |
})
|
| 30 |
+
# Nested dict, e.g., user: { username: ... }
|
| 31 |
+
if k in COMMON_NAME_KEYS and isinstance(v, dict):
|
| 32 |
+
for nk, nv in v.items():
|
| 33 |
+
if nk in COMMON_NAME_KEYS and norm_target in normalize(nv):
|
| 34 |
+
login_info = {}
|
| 35 |
+
for lk in LOGIN_KEYS:
|
| 36 |
+
if lk in obj:
|
| 37 |
+
login_info[lk] = obj[lk]
|
| 38 |
+
matches.append({
|
| 39 |
+
"match_path": path + [k, nk],
|
| 40 |
+
"matched_name": nv,
|
| 41 |
+
"record": obj,
|
| 42 |
+
"file": file_name,
|
| 43 |
+
"login_info": login_info
|
| 44 |
+
})
|
| 45 |
_search(v, path + [k])
|
| 46 |
elif isinstance(obj, list):
|
| 47 |
for idx, item in enumerate(obj):
|
|
|
|
| 49 |
_search(data, [])
|
| 50 |
return matches
|
| 51 |
|
|
|
|
| 52 |
def show_all_users():
|
| 53 |
found = []
|
| 54 |
for file_name, data in st.session_state.json_data.items():
|
|
|
|
| 56 |
if isinstance(obj, dict):
|
| 57 |
for k, v in obj.items():
|
| 58 |
if k in COMMON_NAME_KEYS:
|
| 59 |
+
if isinstance(v, str):
|
| 60 |
+
found.append(f"{file_name} | {'.'.join(path + [k])} = {v}")
|
| 61 |
+
elif isinstance(v, dict):
|
| 62 |
+
for nk, nv in v.items():
|
| 63 |
+
if nk in COMMON_NAME_KEYS:
|
| 64 |
+
found.append(f"{file_name} | {'.'.join(path + [k, nk])} = {nv}")
|
| 65 |
recursive(v, path + [k])
|
| 66 |
elif isinstance(obj, list):
|
| 67 |
for idx, item in enumerate(obj):
|
|
|
|
| 69 |
recursive(data, [])
|
| 70 |
return found
|
| 71 |
|
|
|
|
| 72 |
import re
|
| 73 |
def handle_user_query(query):
|
|
|
|
| 74 |
patterns = [
|
| 75 |
r"(?:last\s*login.*?for|when\s+did)\s+([a-zA-Z0-9 _\-\.]+)",
|
| 76 |
r"when\s+was\s+([a-zA-Z0-9 _\-\.]+)\s+last\s+(?:login|logged\s*in)",
|
|
|
|
| 84 |
found_name = m.group(1).strip()
|
| 85 |
break
|
| 86 |
if not found_name:
|
|
|
|
| 87 |
m = re.search(r"([A-Z][a-z]+(?: [A-Z][a-z]+)*)", query)
|
| 88 |
if m:
|
| 89 |
found_name = m.group(1).strip()
|
|
|
|
| 101 |
else:
|
| 102 |
return "Sorry, I can only answer direct user info queries (e.g., 'When did Bob the Builder last login?')."
|
| 103 |
|
| 104 |
+
# --- Streamlit UI ---
|
| 105 |
+
if "json_data" not in st.session_state:
|
| 106 |
+
st.session_state.json_data = {}
|
| 107 |
+
if "messages" not in st.session_state:
|
| 108 |
+
st.session_state.messages = []
|
| 109 |
+
if "temp_input" not in st.session_state:
|
| 110 |
+
st.session_state.temp_input = ""
|
| 111 |
+
if "files_loaded" not in st.session_state:
|
| 112 |
+
st.session_state.files_loaded = False
|
| 113 |
+
|
| 114 |
+
st.set_page_config(page_title="Instant JSON Q&A", layout="wide")
|
| 115 |
+
st.title("Instant JSON-Backed AI Q&A (Now Handles Nested Users!)")
|
| 116 |
+
|
| 117 |
+
uploaded_files = st.sidebar.file_uploader(
|
| 118 |
+
"Choose one or more JSON files", type="json", accept_multiple_files=True
|
| 119 |
+
)
|
| 120 |
+
if uploaded_files and not st.session_state.files_loaded:
|
| 121 |
+
st.session_state.json_data.clear()
|
| 122 |
+
for f in uploaded_files:
|
| 123 |
+
try:
|
| 124 |
+
content = json.load(f)
|
| 125 |
+
st.session_state.json_data[f.name] = content
|
| 126 |
+
st.sidebar.success(f"Loaded: {f.name}")
|
| 127 |
+
except Exception as e:
|
| 128 |
+
st.sidebar.error(f"Error reading {f.name}: {e}")
|
| 129 |
+
st.session_state.messages = []
|
| 130 |
+
st.session_state.files_loaded = True
|
| 131 |
+
elif not uploaded_files:
|
| 132 |
+
st.session_state.json_data.clear()
|
| 133 |
+
st.session_state.files_loaded = False
|
| 134 |
+
|
| 135 |
st.markdown("### Ask about any user directly (e.g. 'When did Bob the Builder last login?')")
|
| 136 |
for msg in st.session_state.messages:
|
| 137 |
if msg["role"] == "user":
|