File size: 10,131 Bytes
e212866
 
 
 
388e3b5
e212866
388e3b5
 
 
 
8568540
388e3b5
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
8568540
 
e212866
 
 
 
 
 
 
 
 
8568540
 
e212866
 
8568540
 
e212866
 
 
 
 
8568540
 
 
 
 
 
 
e212866
 
 
 
 
 
 
 
 
 
 
 
 
8568540
e212866
 
8568540
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
e212866
 
 
 
 
 
 
388e3b5
e212866
8568540
e212866
8568540
 
388e3b5
 
 
 
 
 
 
 
 
 
8568540
388e3b5
 
8568540
 
 
 
 
 
 
 
 
 
 
e212866
8568540
388e3b5
e212866
 
8568540
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
388e3b5
8568540
 
e212866
 
 
 
8568540
e212866
 
 
 
 
8568540
 
 
e212866
8568540
 
 
e212866
 
 
8568540
 
e212866
8568540
 
 
 
 
e212866
8568540
 
e212866
8568540
e212866
 
 
8568540
 
e212866
 
8568540
 
 
 
 
e212866
 
 
 
 
 
 
388e3b5
e212866
 
 
 
 
 
 
 
2b37229
e212866
 
 
2b37229
 
 
 
 
 
 
 
 
 
 
 
e212866
8568540
 
 
2b37229
e212866
8568540
 
e212866
8568540
 
 
e212866
 
 
 
 
 
 
8568540
e212866
 
 
 
8568540
 
 
 
 
 
 
 
e212866
 
 
 
8568540
 
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
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
import sqlite3
import json
import uuid
import datetime
import logging

# Setup logging
logging.basicConfig(level=logging.INFO, format='%(asctime)s - %(levelname)s - %(message)s')

# Attempt to import transformers
try:
    from transformers import AutoModelForCausalLM, AutoTokenizer
    TRANSFORMERS_AVAILABLE = True
except ImportError:
    logging.warning("Transformers library not found. Using fallback parser.")
    TRANSFORMERS_AVAILABLE = False
    AutoModelForCausalLM = None
    AutoTokenizer = None

# Initialize AI model (distilbert as placeholder; replace with fine-tuned model or Mistral-7B)
model_name = "distilbert-base-uncased"  # Lightweight model for demo
if TRANSFORMERS_AVAILABLE:
    try:
        tokenizer = AutoTokenizer.from_pretrained(model_name)
        model = AutoModelForCausalLM.from_pretrained(model_name)
        logging.info(f"Loaded model: {model_name}")
    except Exception as e:
        logging.error(f"Failed to load model {model_name}: {e}")
        tokenizer = None
        model = None
else:
    tokenizer = None
    model = None

# Database setup
conn = sqlite3.connect("erp.db")
cursor = conn.cursor()

# Create tables
cursor.execute("""
CREATE TABLE IF NOT EXISTS chart_of_accounts (
    account_id TEXT PRIMARY KEY,
    account_name TEXT NOT NULL,
    account_type TEXT NOT NULL,
    parent_id TEXT,
    allow_budgeting BOOLEAN,
    allow_posting BOOLEAN,
    FOREIGN KEY (parent_id) REFERENCES chart_of_accounts(account_id)
)
""")
cursor.execute("""
CREATE TABLE IF NOT EXISTS journal_entries (
    entry_id TEXT PRIMARY KEY,
    date TEXT NOT NULL,
    debit_account_id TEXT NOT NULL,
    credit_account_id TEXT NOT NULL,
    amount REAL NOT NULL,
    description TEXT,
    FOREIGN KEY (debit_account_id) REFERENCES chart_of_accounts(account_id),
    FOREIGN KEY (credit_account_id) REFERENCES chart_of_accounts(account_id)
)
""")
conn.commit()

# Debit/Credit rules
ACCOUNT_RULES = {
    "Asset": {"increase": "Debit", "decrease": "Credit"},
    "Liability": {"increase": "Credit", "decrease": "Debit"},
    "Equity": {"increase": "Credit", "decrease": "Debit"},
    "Revenue": {"increase": "Credit", "decrease": "Debit"},
    "Expense": {"increase": "Debit", "decrease": "Credit"}
}

# Initialize chart of accounts
def initialize_chart_of_accounts():
    accounts = [
        ("1", "Assets", "Asset", None, True, False),
        ("1.1", "Fixed Assets", "Asset", "1", True, False),
        ("1.1.1", "Plant", "Asset", "1.1", True, True),
        ("1.1.2", "Machinery", "Asset", "1.1", True, True),
        ("1.1.3", "Building", "Asset", "1.1", True, True),
        ("1.2", "Current Assets", "Asset", "1", True, False),
        ("1.2.1", "Cash", "Asset", "1.2", True, True),
        ("1.2.2", "Laptop", "Asset", "1.2", True, True),
        ("2", "Liabilities", "Liability", None, True, False),
        ("2.1", "Accounts Payable", "Liability", "2", True, True),
        ("3", "Equity", "Equity", None, True, False),
        ("3.1", "Owner's Capital", "Equity", "3", True, True),
        ("4", "Revenue", "Revenue", None, True, False),
        ("4.1", "Sales", "Revenue", "4", True, True),
        ("5", "Expenses", "Expense", None, True, False),
        ("5.1", "Operating Expenses", "Expense", "5", True, True)
    ]
    cursor.executemany("""
    INSERT OR REPLACE INTO chart_of_accounts
    (account_id, account_name, account_type, parent_id, allow_budgeting, allow_posting)
    VALUES (?, ?, ?, ?, ?, ?)
    """, accounts)
    conn.commit()
    logging.info("Chart of accounts initialized.")

# Parse prompt using AI model (or fallback)
def parse_prompt(prompt):
    if model and tokenizer:
        try:
            input_text = f"""
            Parse the following accounting prompt into a JSON object with:
            - debit: {{account, type, amount}}
            - credit: {{account, type, amount}}
            - payment_method: 'cash' or 'credit' or null
            Prompt: {prompt}
            """
            inputs = tokenizer(input_text, return_tensors="pt")
            outputs = model.generate(**inputs, max_length=300)
            response = tokenizer.decode(outputs[0], skip_special_tokens=True)
            return json.loads(response)
        except Exception as e:
            logging.warning(f"Model parsing failed: {e}. Using fallback parser.")

    # Fallback parsing for common scenarios
    prompt_lower = prompt.lower()
    amount = None
    for word in prompt_lower.split():
        if word.startswith("$"):
            try:
                amount = float(word[1:])
                break
            except:
                pass
    
    if not amount:
        logging.error("No amount found in prompt.")
        return None

    if "laptop" in prompt_lower:
        debit_account = "Laptop"
        debit_type = "Asset"
        if "cash" in prompt_lower:
            credit_account = "Cash"
            credit_type = "Asset"
            payment_method = "cash"
        elif "credit" in prompt_lower:
            credit_account = "Accounts Payable"
            credit_type = "Liability"
            payment_method = "credit"
        else:
            return {"debit": {"account": "Laptop", "type": "Asset", "amount": amount}, "credit": None, "payment_method": None}
        return {
            "debit": {"account": debit_account, "type": debit_type, "amount": amount},
            "credit": {"account": credit_account, "type": credit_type, "amount": amount},
            "payment_method": payment_method
        }
    logging.error("Prompt not recognized.")
    return None

# Generate journal entry
def generate_journal_entry(prompt, follow_up_response=None):
    parsed = parse_prompt(prompt)
    if not parsed:
        return "Unable to parse prompt. Please provide more details."

    debit_account = parsed["debit"]["account"]
    amount = parsed["debit"]["amount"]
    payment_method = parsed.get("payment_method")

    # Handle ambiguous payment method
    if not payment_method and not follow_up_response:
        return {"status": "clarify", "message": "Wonderful, did you buy on credit? (Yes/No)"}

    # Determine credit account
    credit_account = None
    credit_type = None
    if follow_up_response and follow_up_response.lower() == "yes":
        credit_account = "Accounts Payable"
        credit_type = "Liability"
    elif payment_method == "cash":
        credit_account = parsed["credit"]["account"]
        credit_type = parsed["credit"]["type"]
    elif payment_method == "credit":
        credit_account = "Accounts Payable"
        credit_type = "Liability"
    else:
        return "Invalid payment method specified."

    # Validate accounts
    cursor.execute("SELECT account_id, account_type, allow_posting FROM chart_of_accounts WHERE account_name = ?", (debit_account,))
    debit_result = cursor.fetchone()
    cursor.execute("SELECT account_id, account_type, allow_posting FROM chart_of_accounts WHERE account_name = ?", (credit_account,))
    credit_result = cursor.fetchone()

    if not debit_result or not credit_result:
        return "One or both accounts not found in chart of accounts."
    if not debit_result[2] or not credit_result[2]:
        return "Posting not allowed for one or both accounts."

    # Validate account types
    if debit_result[1] != parsed["debit"]["type"] or credit_result[1] != credit_type:
        return "Account type mismatch."

    # Create journal entry
    entry_id = str(uuid.uuid4())
    date = datetime.datetime.now().isoformat()
    cursor.execute("""
    INSERT INTO journal_entries (entry_id, date, debit_account_id, credit_account_id, amount, description)
    VALUES (?, ?, ?, ?, ?, ?)
    """, (entry_id, date, debit_result[0], credit_result[0], amount, prompt))
    conn.commit()
    logging.info(f"Journal entry created: Debit {debit_account} ${amount}, Credit {credit_account} ${amount}")

    return f"Journal Entry Created: Debit {debit_account} ${amount}, Credit {credit_account} ${amount}"

# Generate T-account
def generate_t_account(account_name):
    cursor.execute("SELECT account_id FROM chart_of_accounts WHERE account_name = ?", (account_name,))
    account_id = cursor.fetchone()
    if not account_id:
        logging.error(f"Account {account_name} not found.")
        return "Account not found."
    
    account_id = account_id[0]
    try:
        cursor.execute("""
        SELECT date, amount, description, 'Debit' as type FROM journal_entries WHERE debit_account_id = ?
        UNION
        SELECT date, amount, description, 'Credit' as type FROM journal_entries WHERE credit_account_id = ?
        ORDER BY date
        """, (account_id, account_id))
        entries = cursor.fetchall()
        logging.info(f"Retrieved {len(entries)} entries for T-account: {account_name}")
    except sqlite3.Error as e:
        logging.error(f"SQL error in generate_t_account: {e}")
        return "Error retrieving T-account data."
    
    t_account = f"T-Account for {account_name}\n{'='*50}\n{'Debit':<20} | {'Credit':<20} | Description\n{'-'*50}\n"
    debit_total = 0
    credit_total = 0
    for date, amount, desc, entry_type in entries:
        if entry_type == "Debit":
            t_account += f"${amount:<19} | {'':<20} | {desc}\n"
            debit_total += amount
        else:
            t_account += f"{'':<20} | ${amount:<19} | {desc}\n"
            credit_total += amount
    t_account += f"{'-'*50}\nTotal Debit: ${debit_total:<10} | Total Credit: ${credit_total}\n"
    
    return t_account

# Example usage
if __name__ == "__main__":
    initialize_chart_of_accounts()
    
    # Test prompt (cash purchase)
    prompt = "Bought a laptop for $200 on cash"
    result = generate_journal_entry(prompt)
    print(result)
    
    # Test prompt (ambiguous payment method)
    prompt = "Bought a laptop for $300"
    result = generate_journal_entry(prompt)
    print(result)
    if isinstance(result, dict) and result["status"] == "clarify":
        result = generate_journal_entry(prompt, "Yes")
        print(result)
    
    # Test T-account
    t_account = generate_t_account("Laptop")
    print(t_account)
    
    # Clean up
    conn.close()