Files
baiyue-logistics-chat/logistics_chat/quote_data.py
T

145 lines
7.6 KiB
Python

"""Versioned, private quote imports and read-only runtime access."""
import json
import re
import sqlite3
from decimal import Decimal, InvalidOperation
def init_quote_tables(connection):
connection.execute('''CREATE TABLE IF NOT EXISTS quote_books (
id INTEGER PRIMARY KEY, source_sha256 TEXT NOT NULL UNIQUE,
source_name TEXT NOT NULL, active INTEGER NOT NULL DEFAULT 0)''')
connection.execute('''CREATE TABLE IF NOT EXISTS quote_rates (
id INTEGER PRIMARY KEY, book_id INTEGER NOT NULL, service TEXT NOT NULL,
kind TEXT NOT NULL, origin TEXT NOT NULL, zone TEXT NOT NULL,
unit TEXT NOT NULL, tax TEXT NOT NULL, min_weight TEXT NOT NULL,
price TEXT NOT NULL, eta TEXT NOT NULL, destinations_json TEXT NOT NULL,
source_sheet TEXT NOT NULL, source_cell TEXT NOT NULL)''')
connection.execute('CREATE INDEX IF NOT EXISTS idx_quote_rates_book ON quote_rates(book_id)')
connection.execute('''CREATE TABLE IF NOT EXISTS quote_warehouses (
id INTEGER PRIMARY KEY, book_id INTEGER NOT NULL, code TEXT NOT NULL,
zone TEXT NOT NULL, state TEXT NOT NULL, postal TEXT NOT NULL,
address TEXT NOT NULL, source_sheet TEXT NOT NULL, source_cell TEXT NOT NULL)''')
connection.execute('CREATE INDEX IF NOT EXISTS idx_quote_warehouse_code ON quote_warehouses(book_id,code)')
connection.execute('''CREATE TABLE IF NOT EXISTS quote_postcodes (
book_id INTEGER NOT NULL, postal TEXT NOT NULL, severity TEXT NOT NULL,
fee TEXT, PRIMARY KEY(book_id,postal))''')
connection.execute('''CREATE TABLE IF NOT EXISTS quote_documents (
id INTEGER PRIMARY KEY, book_id INTEGER NOT NULL, sheet TEXT NOT NULL,
source_cell TEXT NOT NULL, content TEXT NOT NULL)''')
def decimal_text(value, positive=False):
try:
number = Decimal(str(value))
except (InvalidOperation, ValueError):
raise ValueError('Invalid quote amount') from None
if not number.is_finite() or number < 0 or number > 100000000 or (positive and not number):
raise ValueError('Invalid quote amount')
return format(number, 'f')
def import_snapshot(connection, snapshot, *, activate=False):
"""Validate before writing; replace an active book atomically, never order data."""
if not isinstance(snapshot, dict) or snapshot.get('schema') != 1:
raise ValueError('Unsupported quote snapshot')
digest = snapshot.get('source_sha256', '')
if not re.fullmatch('[0-9a-f]{64}', digest):
raise ValueError('Invalid quote source checksum')
source_name = str(snapshot.get('source_name', ''))[:240]
if not source_name:
raise ValueError('Missing quote source name')
limits = {'rates':10000, 'warehouses':10000, 'postcodes':50000, 'documents':5000}
for key, limit in limits.items():
if not isinstance(snapshot.get(key), list) or len(snapshot[key]) > limit:
raise ValueError('Invalid quote records')
rates = []
for row in snapshot['rates']:
if (row.get('kind') not in ('truck', 'parcel') or row.get('unit') not in ('kg', 'cbm')
or row.get('tax') not in ('included', 'excluded') or not row.get('service')
or not isinstance(row.get('destinations'), list)
or len(row['destinations']) > 200):
raise ValueError('Invalid quote rate')
rates.append((str(row['service'])[:100], row['kind'], str(row.get('origin', ''))[:40],
str(row.get('zone', ''))[:40], row['unit'], row['tax'],
decimal_text(row.get('min_weight', 0)), decimal_text(row['price'], positive=True),
str(row.get('eta', ''))[:200], json.dumps(row['destinations'], ensure_ascii=False),
str(row['source_sheet'])[:100], str(row['source_cell'])[:20]))
warehouses = [(str(r.get('code', '')).strip().upper()[:40], str(r.get('zone', ''))[:40],
str(r.get('state', ''))[:40], str(r.get('postal', ''))[:20], str(r.get('address', ''))[:1000],
str(r['source_sheet'])[:100], str(r['source_cell'])[:20]) for r in snapshot['warehouses']]
postcodes = []
for row in snapshot['postcodes']:
if not re.fullmatch('[0-9]{5}', row.get('postal', '')) or row.get('severity') not in ('remote','extended','extreme'):
raise ValueError('Invalid quote postcode')
postcodes.append((row['postal'], row['severity'], None if row.get('fee') is None else decimal_text(row['fee'])))
documents = [(str(r['sheet'])[:100], str(r['cell'])[:20], str(r['text'])[:16000]) for r in snapshot['documents']]
if not rates:
raise ValueError('Quote snapshot contains no rates')
init_quote_tables(connection)
if not connection.in_transaction:
connection.execute('BEGIN IMMEDIATE')
connection.execute('SAVEPOINT quote_import')
try:
existing = connection.execute('SELECT id FROM quote_books WHERE source_sha256=?', (digest,)).fetchone()
if existing:
book_id = existing[0]
else:
book_id = connection.execute('INSERT INTO quote_books(source_sha256,source_name) VALUES(?,?)',
(digest, source_name)).lastrowid
connection.executemany('''INSERT INTO quote_rates(book_id,service,kind,origin,zone,unit,tax,
min_weight,price,eta,destinations_json,source_sheet,source_cell) VALUES(?,?,?,?,?,?,?,?,?,?,?,?,?)''',
[(book_id, *r) for r in rates])
connection.executemany('''INSERT INTO quote_warehouses(book_id,code,zone,state,postal,address,source_sheet,source_cell)
VALUES(?,?,?,?,?,?,?,?)''', [(book_id, *r) for r in warehouses])
connection.executemany('INSERT INTO quote_postcodes(book_id,postal,severity,fee) VALUES(?,?,?,?)',
[(book_id, *r) for r in postcodes])
connection.executemany('INSERT INTO quote_documents(book_id,sheet,source_cell,content) VALUES(?,?,?,?)',
[(book_id, *r) for r in documents])
if activate:
connection.execute('UPDATE quote_books SET active=0 WHERE active<>0')
connection.execute('UPDATE quote_books SET active=1 WHERE id=?', (book_id,))
connection.execute('RELEASE quote_import')
except Exception:
connection.execute('ROLLBACK TO quote_import')
connection.execute('RELEASE quote_import')
raise
return {'book_id':book_id, 'rates':len(rates), 'warehouses':len(warehouses),
'postcodes':len(postcodes), 'documents':len(documents), 'active':bool(activate), 'existing':bool(existing)}
def load_book():
"""One read transaction prevents mixed versions during a concurrent import."""
from .db import connect
connection = connect()
try:
connection.row_factory = sqlite3.Row
connection.execute('BEGIN')
book = connection.execute('SELECT id FROM quote_books WHERE active=1 ORDER BY id DESC LIMIT 1').fetchone()
if not book:
return None
book_id = book['id']
result = {'id':book_id}
for table, name in (('quote_rates','rates'), ('quote_warehouses','warehouses'), ('quote_documents','documents')):
result[name] = [dict(r) for r in connection.execute(f'SELECT * FROM {table} WHERE book_id=?', (book_id,))]
for rate in result['rates']:
rate['destinations'] = json.loads(rate.pop('destinations_json'))
return result
finally:
connection.close()
def postcode_fee(book_id, postal):
from .db import connect
connection = connect()
try:
connection.row_factory = sqlite3.Row
row = connection.execute('SELECT severity,fee FROM quote_postcodes WHERE book_id=? AND postal=?',
(book_id, postal)).fetchone()
return dict(row) if row else None
finally:
connection.close()