Spaces:
Sleeping
Sleeping
Intention commited on
Commit Β·
b6b1f02
1
Parent(s): fdf8278
edit pii.
Browse files
app.py
CHANGED
|
@@ -9,10 +9,11 @@ from datetime import datetime
|
|
| 9 |
from uuid import uuid4
|
| 10 |
|
| 11 |
# -----------------------------
|
| 12 |
-
#
|
| 13 |
# -----------------------------
|
| 14 |
st.set_page_config(page_title="ChatGPT Log Analyzer", page_icon="π€")
|
| 15 |
|
|
|
|
| 16 |
if "consent" not in st.session_state:
|
| 17 |
st.session_state.consent = ""
|
| 18 |
|
|
@@ -40,11 +41,21 @@ if st.session_state.consent == "Yes, I consent":
|
|
| 40 |
st.session_state.id = datetime.now().strftime('%Y%m-%d%H-%M-') + str(uuid4())
|
| 41 |
st.success("β
You consented to participate.")
|
| 42 |
st.info(f"Your anonymized ID is: **{st.session_state.id}**. Keep this if you want your data deleted later.")
|
| 43 |
-
|
| 44 |
elif st.session_state.consent == "No, I do not consent":
|
| 45 |
placeholder.empty()
|
| 46 |
st.warning("β οΈ You did not consent. You can still use the app, but your logs will not be stored.")
|
| 47 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 48 |
# -----------------------------
|
| 49 |
# Parser Function
|
| 50 |
# -----------------------------
|
|
@@ -82,39 +93,17 @@ if uploaded_file:
|
|
| 82 |
data = json.load(uploaded_file)
|
| 83 |
if isinstance(data, dict) and "conversations" in data:
|
| 84 |
df = parse_chatgpt_export(data)
|
| 85 |
-
convo_titles = df["title"].unique()
|
| 86 |
else:
|
| 87 |
st.error("Unsupported JSON structure")
|
| 88 |
st.stop()
|
| 89 |
|
| 90 |
# -----------------------------
|
| 91 |
-
#
|
| 92 |
-
# -----------------------------
|
| 93 |
-
if "convo_index" not in st.session_state:
|
| 94 |
-
st.session_state.convo_index = 0
|
| 95 |
-
|
| 96 |
-
# -----------------------------
|
| 97 |
-
# Dropdown Selection
|
| 98 |
# -----------------------------
|
| 99 |
-
|
| 100 |
-
|
|
|
|
| 101 |
|
| 102 |
-
# -----------------------------
|
| 103 |
-
# Pagination Buttons
|
| 104 |
-
# -----------------------------
|
| 105 |
-
col1, col2, col3 = st.columns([1,2,1])
|
| 106 |
-
with col1:
|
| 107 |
-
if st.button("β¬
Previous") and st.session_state.convo_index > 0:
|
| 108 |
-
st.session_state.convo_index -= 1
|
| 109 |
-
selected_title = convo_titles[st.session_state.convo_index]
|
| 110 |
-
with col3:
|
| 111 |
-
if st.button("Next β‘") and st.session_state.convo_index < len(convo_titles)-1:
|
| 112 |
-
st.session_state.convo_index += 1
|
| 113 |
-
selected_title = convo_titles[st.session_state.convo_index]
|
| 114 |
-
|
| 115 |
-
# -----------------------------
|
| 116 |
-
# Filter Selected Conversation
|
| 117 |
-
# -----------------------------
|
| 118 |
convo_df = df[df["title"] == selected_title].copy()
|
| 119 |
|
| 120 |
# -----------------------------
|
|
@@ -127,9 +116,23 @@ if uploaded_file:
|
|
| 127 |
convo_df["sentiment"] = convo_df["content"].apply(lambda x: analyzer.polarity_scores(str(x))["compound"])
|
| 128 |
|
| 129 |
# -----------------------------
|
| 130 |
-
#
|
| 131 |
# -----------------------------
|
| 132 |
-
st.subheader(f"π¬ Conversation
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 133 |
st.dataframe(convo_df[["role", "redacted", "sentiment", "create_time"]])
|
| 134 |
|
| 135 |
# -----------------------------
|
|
@@ -146,4 +149,4 @@ if uploaded_file:
|
|
| 146 |
"messages": convo_df.to_dict(orient="records")
|
| 147 |
}
|
| 148 |
collection.insert_one(record)
|
| 149 |
-
st.success(f"β
Conversation '{selected_title}' saved to MongoDB.")
|
|
|
|
| 9 |
from uuid import uuid4
|
| 10 |
|
| 11 |
# -----------------------------
|
| 12 |
+
# Page Config & Consent
|
| 13 |
# -----------------------------
|
| 14 |
st.set_page_config(page_title="ChatGPT Log Analyzer", page_icon="π€")
|
| 15 |
|
| 16 |
+
# Consent
|
| 17 |
if "consent" not in st.session_state:
|
| 18 |
st.session_state.consent = ""
|
| 19 |
|
|
|
|
| 41 |
st.session_state.id = datetime.now().strftime('%Y%m-%d%H-%M-') + str(uuid4())
|
| 42 |
st.success("β
You consented to participate.")
|
| 43 |
st.info(f"Your anonymized ID is: **{st.session_state.id}**. Keep this if you want your data deleted later.")
|
|
|
|
| 44 |
elif st.session_state.consent == "No, I do not consent":
|
| 45 |
placeholder.empty()
|
| 46 |
st.warning("β οΈ You did not consent. You can still use the app, but your logs will not be stored.")
|
| 47 |
|
| 48 |
+
# -----------------------------
|
| 49 |
+
# Privacy Policy Dropdown
|
| 50 |
+
# -----------------------------
|
| 51 |
+
with st.expander("Privacy Policy", expanded=False):
|
| 52 |
+
st.markdown("""
|
| 53 |
+
**Data Collection:** Only ChatGPT logs you upload will be analyzed.
|
| 54 |
+
**PII Redaction:** All detected names, emails, and other personal info will be redacted.
|
| 55 |
+
**Storage:** Data is stored anonymized in our research database.
|
| 56 |
+
**Deletion:** You can request deletion using your anonymized ID.
|
| 57 |
+
""")
|
| 58 |
+
|
| 59 |
# -----------------------------
|
| 60 |
# Parser Function
|
| 61 |
# -----------------------------
|
|
|
|
| 93 |
data = json.load(uploaded_file)
|
| 94 |
if isinstance(data, dict) and "conversations" in data:
|
| 95 |
df = parse_chatgpt_export(data)
|
|
|
|
| 96 |
else:
|
| 97 |
st.error("Unsupported JSON structure")
|
| 98 |
st.stop()
|
| 99 |
|
| 100 |
# -----------------------------
|
| 101 |
+
# Conversation Selector
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 102 |
# -----------------------------
|
| 103 |
+
st.subheader("π Select a Conversation")
|
| 104 |
+
convo_titles = df["title"].unique()
|
| 105 |
+
selected_title = st.selectbox("Choose conversation", convo_titles)
|
| 106 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 107 |
convo_df = df[df["title"] == selected_title].copy()
|
| 108 |
|
| 109 |
# -----------------------------
|
|
|
|
| 116 |
convo_df["sentiment"] = convo_df["content"].apply(lambda x: analyzer.polarity_scores(str(x))["compound"])
|
| 117 |
|
| 118 |
# -----------------------------
|
| 119 |
+
# Inline PII Editing
|
| 120 |
# -----------------------------
|
| 121 |
+
st.subheader(f"π¬ Conversation: {selected_title}")
|
| 122 |
+
editable_rows = []
|
| 123 |
+
for i, row in convo_df.iterrows():
|
| 124 |
+
st.markdown(f"**{row['role'].capitalize()} ({row['create_time']}):**")
|
| 125 |
+
edited_text = st.text_area(
|
| 126 |
+
f"Message {i}",
|
| 127 |
+
value=row["redacted"],
|
| 128 |
+
key=f"edit_{i}"
|
| 129 |
+
)
|
| 130 |
+
editable_rows.append({
|
| 131 |
+
**row,
|
| 132 |
+
"redacted": edited_text
|
| 133 |
+
})
|
| 134 |
+
|
| 135 |
+
convo_df = pd.DataFrame(editable_rows)
|
| 136 |
st.dataframe(convo_df[["role", "redacted", "sentiment", "create_time"]])
|
| 137 |
|
| 138 |
# -----------------------------
|
|
|
|
| 149 |
"messages": convo_df.to_dict(orient="records")
|
| 150 |
}
|
| 151 |
collection.insert_one(record)
|
| 152 |
+
st.success(f"β
Conversation '{selected_title}' saved to MongoDB.")
|