File size: 9,094 Bytes
4d86911
 
 
30a77ea
4d86911
 
 
 
 
 
 
 
 
 
3b1b499
4d86911
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
30a77ea
4d86911
 
 
 
 
30a77ea
4d86911
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
30a77ea
4d86911
 
30a77ea
4d86911
 
 
 
 
 
30a77ea
4d86911
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
3b1b499
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
4d86911
3b1b499
4d86911
3b1b499
 
 
4d86911
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
30a77ea
4d86911
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
# Fixed and Hugging Face Spaces-Compatible Code

import os
import streamlit as st
import pandas as pd
import subprocess
import json
import plotly.express as px
import re
import io
import requests
from sqlalchemy import create_engine, text, inspect

# --- Get HF Token ---
HF_TOKEN = os.environ.get("HF_TOKEN", "")  # Safely get token, fallback if missing

# --- Helper: Call Mistral Model ---
def mistral_call(schema=None, question="no questions were asked", hf_token=HF_TOKEN, model_id="mistralai/Mistral-7B-Instruct-v0.3"):
    api_url = f"https://api-inference.huggingface.co/models/{model_id}"
    headers = {
        "Authorization": f"Bearer {hf_token}",
        "Content-Type": "application/json"
    }
    prompt = f"""You are a helpful assistant that translates natural language questions into SQL using a database schema.
### Schema:
{schema}
### Question:
{question}
"""
    payload = {
        "inputs": prompt,
        "parameters": {
            "max_new_tokens": 500,
            "do_sample": True,
            "temperature": 0.3,
        }
    }
    response = requests.post(api_url, headers=headers, json=payload)
    if response.status_code == 200:
        try:
            generated = response.json()[0]['generated_text']
            return generated.split("### Question:")[-1].strip()
        except Exception as e:
            return f"Error parsing response: {e}"
    else:
        return f"API call failed: {response.status_code}\n{response.text}"

# --- Visualization Suggestion ---
def extract_json(text):
    match = re.search(r"\{.*?\}", text, re.DOTALL)
    if match:
        try:
            return json.loads(match.group(0))
        except json.JSONDecodeError:
            return None
    return None

def get_visualization_suggestion(data):
    prompt = f"""
These are the dataset column names: {list(data.columns)}.
Suggest one visualization using the format:
{{"x": "column", "y": "column or list", "chart_type": "bar/line/scatter/pie"}}
"""
    response = mistral_call(question=prompt)
    return extract_json(response)

# --- Demo Data Generator ---
def generate_demo_data_csv(user_input, num_rows=10):
    prompt = f"""
Generate a {num_rows}-row structured dataset in CSV format with quoted column headers and values:
"{user_input}"
"""
    response = mistral_call(question=prompt)
    csv_data = "\n".join([line.strip() for line in response.splitlines() if line.strip().startswith('"')])
    if csv_data:
        try:
            df = pd.read_csv(io.StringIO(csv_data))
            buffer = io.StringIO()
            df.to_csv(buffer, index=False)
            return "Demo data generated.", buffer
        except Exception as e:
            return f"CSV error: {e}", None
    return "No CSV found.", None

# --- SQL Utilities ---
def extract_sql_code_blocks(text):
    return re.findall(r"```sql\s+(.*?)```", text, re.DOTALL | re.IGNORECASE)

def remove_think_tags(text):
    return re.sub(r'<think>.*?</think>', '', text, flags=re.DOTALL | re.IGNORECASE)

def classify_sql_task_prompt_engineered(user_input: str) -> str:
    prompt = f"""
Classify into:
CREATE_TABLE, INSERT_INTO, SELECT, UPDATE, DELETE, ALTER_TABLE, INSERT_CSV_EXISTING, INSERT_CSV_NEW
Input: {user_input}
Only return the task.
"""
    classification = mistral_call(question=prompt)
    cleaned = remove_think_tags(classification).strip().upper()
    for t in ["CREATE_TABLE", "INSERT_INTO", "SELECT", "UPDATE", "DELETE", "ALTER_TABLE", "INSERT_CSV_EXISTING", "INSERT_CSV_NEW"]:
        if t in cleaned:
            return t
    return "UNKNOWN"

def handle_query(user_input, engine, task_type):
    try:
        inspector = inspect(engine)
        tables = inspector.get_table_names()
        prompt = f"Generate {task_type} SQL for: {user_input} using tables: {tables}"
        sql_code = mistral_call(question=prompt)
        sql_code = extract_sql_code_blocks(sql_code)
        return execute_sql(sql_code, engine)
    except Exception as e:
        return "None", f"Error: {e}"

def execute_sql(sql_code, engine):
    try:
        if isinstance(sql_code, list):
            sql_code = "\n".join(sql_code)
        statements = [stmt.strip() for stmt in sql_code.split(';') if stmt.strip()]
        with engine.connect() as conn:
            for stmt in statements:
                conn.execute(text(stmt + ";"))
            conn.commit()
        return sql_code, "βœ… SQL executed."
    except Exception as e:
        return "None", f"SQL error: {e}"

def insert_csv_existing(table_name, csv_file, engine):
    try:
        df = pd.read_csv(csv_file)
        df.to_sql(table_name, engine, if_exists='append', index=False)
        return f"βœ… CSV inserted into '{table_name}'."
    except Exception as e:
        return f"CSV insert error: {e}"

def insert_csv_new(table_name, csv_file, engine):
    try:
        df = pd.read_csv(csv_file)
        df.to_sql(table_name, engine, if_exists='replace', index=False)
        return f"βœ… CSV inserted into new table '{table_name}'."
    except Exception as e:
        return f"New CSV insert error: {e}"

# --- Streamlit App ---
st.set_page_config(page_title="AI Dashboard", layout="wide")
st.title("πŸ€– AI-Powered Multi-Feature Dashboard")

st.sidebar.title("Navigation")
option = st.sidebar.radio("Select Feature", ["πŸ“Š Data Visualization", "🧠 SQL Query Generator", "πŸ“„ Demo Data Generator", "🧠 Smart SQL Task Handler"])

if option == "πŸ“Š Data Visualization":
    uploaded_file = st.file_uploader("Upload your CSV", type="csv")
    if uploaded_file:
        try:
            content = uploaded_file.getvalue().decode("utf-8")
            df = pd.read_csv(io.StringIO(content))
            df.columns = df.columns.str.strip().str.replace(" ", "_")
            st.write("CSV Preview")
            st.dataframe(df.head())
            st.write("Shape:", df.shape)

            with st.spinner("Getting chart suggestion..."):
                suggestion = get_visualization_suggestion(df)

            st.write("Model suggestion:")
            st.code(suggestion)

            if suggestion:
                x_col = suggestion.get("x", "").strip()
                y_col = suggestion.get("y", [])
                y_col = [y_col] if isinstance(y_col, str) else y_col
                chart = suggestion.get("chart_type")
                if x_col in df.columns and all(y in df.columns for y in y_col):
                    fig = None
                    if chart == "bar":
                        fig = px.bar(df, x=x_col, y=y_col)
                    elif chart == "line":
                        fig = px.line(df, x=x_col, y=y_col)
                    elif chart == "scatter":
                        fig = px.scatter(df, x=x_col, y=y_col)
                    elif chart == "pie" and len(y_col) == 1:
                        fig = px.pie(df, names=x_col, values=y_col[0])
                    if fig:
                        st.plotly_chart(fig)
                    else:
                        st.error("Unsupported chart type.")
                else:
                    st.error("⚠️ Column suggestion doesn't match your CSV.")
            else:
                st.error("❌ No valid visualization suggestion returned.")
        except Exception as e:
            st.error(f"❌ Error reading CSV: {e}")

elif option == "🧠 SQL Query Generator":
    user_input = st.text_area("Describe your SQL query in plain English:")
    if st.button("Generate SQL"):
        st.code(mistral_call(question=user_input))

elif option == "πŸ“„ Demo Data Generator":
    user_input = st.text_area("Describe your dataset:")
    num_rows = st.number_input("Rows", 1, 1000, 10)
    if st.button("Generate Dataset"):
        msg, buffer = generate_demo_data_csv(user_input, num_rows)
        st.write(msg)
        if buffer:
            st.download_button("Download CSV", buffer.getvalue(), file_name="generated_data.csv", mime="text/csv")

elif option == "🧠 Smart SQL Task Handler":
    st.sidebar.header("DB Settings")
    db_type = "SQLite"
    db_path = st.sidebar.text_input("SQLite File Path", value="smart_sql.db")
    connection_url = f"sqlite:///{db_path}"
    try:
        engine = create_engine(connection_url)
        with engine.connect(): pass
        st.sidebar.success("Connected!")
    except Exception as e:
        st.sidebar.error(f"Connection failed: {e}")
        st.stop()

    user_input = st.text_area("Enter SQL task (or natural language):")
    csv_file = st.file_uploader("Optional CSV Upload")
    table_name = st.text_input("Table name (for CSV):")
    if st.button("Run SQL Task"):
        task = classify_sql_task_prompt_engineered(user_input)
        st.markdown(f"**Detected Task:** `{task}`")
        if task == "INSERT_CSV_EXISTING" and csv_file and table_name:
            st.write(insert_csv_existing(table_name, csv_file, engine))
        elif task == "INSERT_CSV_NEW" and csv_file and table_name:
            st.write(insert_csv_new(table_name, csv_file, engine))
        else:
            sql_code, msg = handle_query(user_input, engine, task)
            st.code(sql_code)
            st.write(msg)