app.py

import streamlit as st
import json
import pandas as pd
import os

st.set_page_config(page_title="Dataset Builder", layout="wide")
st.title("📚 JSONL Dataset Editor")

TMP_DIR = "temp"
TMP_FILE = os.path.join(TMP_DIR, "session_dataset.jsonl")

# --- Helper: ensure tmp dir exists ---
os.makedirs(TMP_DIR, exist_ok=True)


# --- Helper: get all unique fields from records ---
def get_all_fields(data):
    all_keys = set()
    for record in data:
        all_keys.update(record.keys())
    return sorted(all_keys)


# --- Load session data from temp file if exists ---
if "data" not in st.session_state:
    if os.path.exists(TMP_FILE):
        with open(TMP_FILE, "r", encoding="utf-8") as f:
            st.session_state.data = [json.loads(line) for line in f]
        st.session_state.all_fields = get_all_fields(st.session_state.data)
    else:
        st.session_state.data = []
        st.session_state.all_fields = []

# --- Upload JSONL File ---
uploaded_file = st.file_uploader("Upload a JSONL file", type=["jsonl"])

if uploaded_file:
    content = uploaded_file.read().decode("utf-8")
    st.session_state.data = [json.loads(line) for line in content.strip().splitlines()]
    st.session_state.all_fields = get_all_fields(st.session_state.data)

    # Save to temp
    with open(TMP_FILE, "w", encoding="utf-8") as f:
        for item in st.session_state.data:
            f.write(json.dumps(item, ensure_ascii=False) + "\n")

    st.success(
        f"Loaded {len(st.session_state.data)} records with fields: {st.session_state.all_fields}"
    )

# If still no data, use safe fallback fields
if not st.session_state.data and not st.session_state.all_fields:
    st.session_state.all_fields = ["context", "question", "answer"]

# --- Edit Existing Records ---
st.markdown("### ✏️ Edit Records")

df = pd.DataFrame(st.session_state.data)
df = df.reindex(columns=st.session_state.all_fields)

# Fix: Convert likely text fields to string to avoid StreamlitAPIException
for field in st.session_state.all_fields:
    if field.lower() in ["context", "answer", "question"]:
        df[field] = df[field].astype(str)

# Auto-set long fields like "context", "answer" as textareas
column_configs = {
    field: (
        st.column_config.TextColumn(label=field, width="large")
        if field.lower() in ["context", "answer", "question"]
        else None
    )
    for field in st.session_state.all_fields
}

# --- Use st.data_editor for editable table ---
edited_df = st.data_editor(
    df,
    use_container_width=True,
    num_rows="dynamic",
    column_config=column_configs,
)

# --- Save updated data ---
if edited_df is not None:
    st.session_state.data = edited_df.fillna("").to_dict(orient="records")

    # Save to temp file
    with open(TMP_FILE, "w", encoding="utf-8") as f:
        for item in st.session_state.data:
            f.write(json.dumps(item, ensure_ascii=False) + "\n")

    st.toast("✅ Changes auto-saved!", icon="💾")

# --- Add New Entry ---
st.markdown("### ➕ Add New Entry")

# Show form with current fields
with st.form("new_entry_form"):
    new_record = {}
    for field in st.session_state.all_fields:
        new_record[field] = st.text_area(f"{field}", key=f"input_{field}")

    submitted = st.form_submit_button("Add Entry")
    if submitted:
        st.session_state.data.append(new_record)

        # Save to temp
        with open(TMP_FILE, "w", encoding="utf-8") as f:
            for item in st.session_state.data:
                f.write(json.dumps(item, ensure_ascii=False) + "\n")

        st.success("✅ New entry added!")
        st.rerun()

# Option to add a new field
with st.expander("➕ Add New Field"):
    new_field = st.text_input("New field name", key="new_field_name")
    if st.button("Add Field"):
        if new_field and new_field not in st.session_state.all_fields:
            st.session_state.all_fields.append(new_field)
            st.success(f"✅ Field '{new_field}' added!")
            st.rerun()


# --- Export JSONL ---
st.markdown("### 📤 Export Dataset")

# Let user define a custom export path
export_path = st.text_input(
    "Custom save path (e.g., ./exports/my_dataset.jsonl)",
    value="./exports/exported_dataset.jsonl",
)

col1, col2 = st.columns(2)

# --- Export Button ---
with col1:
    if st.button("📁 Export JSONL"):
        if not os.path.exists(os.path.dirname(export_path)):
            os.makedirs(os.path.dirname(export_path))

        # Write to custom path
        with open(export_path, "w", encoding="utf-8") as f_out:
            for row in st.session_state.data:
                f_out.write(json.dumps(row, ensure_ascii=False) + "\n")

        st.success(f"✅ Dataset saved to {export_path}")

        # Load content for download
        with open(export_path, "r", encoding="utf-8") as f_download:
            exported_content = f_download.read()

        st.download_button(
            "⬇️ Download JSONL",
            exported_content,
            file_name=os.path.basename(export_path),
            mime="application/json",
        )

        # Reset session and temp
        if os.path.exists(TMP_FILE):
            os.remove(TMP_FILE)
        st.session_state.clear()
        st.success("🧹 Temporary session cleared. You're starting fresh!")
        st.rerun()

# --- Download Temp Only Button ---
with col2:
    if os.path.exists(TMP_FILE):
        with open(TMP_FILE, "r", encoding="utf-8") as f_tmp:
            tmp_content = f_tmp.read()

        st.download_button(
            "⬇️ Download Temp File",
            tmp_content,
            file_name="session_dataset.jsonl",
            mime="application/json",
        )
    else:
        st.warning("⚠️ No temp file found to download.")