Je, unatumia vipi AI kuandika scripts za uhamiaji wa data na ubadilisho wa muundo?

Question

Accepted Answer

AI ni rahisi sana kwa **ubadilisho wa muundo** (CSV ↔ JSON ↔ SQL) na scripts za uhamiaji kwa sababu sheria za mabadiliko ni kiufundi. Ujanja ni kumpa muktadha wa kutosha kuwa sahihi na salama: schema, safu ya sampuli, na mahitaji ya usalama wazi.

## Jinsi ya kuuliza

- **Toa chanzo na schema ya lengo** — majina ya safu wima, aina, nullability.
- **Jumuisha safu halisi ya sampuli** ili kuona sura halisi na kesi za kingo (manukuu, nulls, tarehe).
- **Omba kuwa idempotent na kuhutimiwa** — kueneza mara nyingine haipaswi kuiga au kusahau data; safu mbaya zinapaswa kukamatwa, si kuingizwa kimya.
- **Daima fanya dry-run kwenye nakala** ya data kabla ya kugusa uzalishaji. Chunguzeni matokeo kwanza.

## Mfano: CSV → SQL inserts

```python
import csv

# Source CSV: id,email,signup_date  (signup_date may be blank)
# Target table: users(id INT, email TEXT NOT NULL, signup_date DATE NULL)
def csv_to_sql(path: str) -> list[str]:
    statements = []
    with open(path, newline="", encoding="utf-8") as f:
        for row in csv.DictReader(f):
            email = row["email"].strip()
            if not email:                      # validate: skip invalid rows, don't insert garbage
                continue
            date = row["signup_date"].strip() or None
            email_sql = email.replace("'", "''")   # escape quotes to avoid broken SQL / injection
            date_sql = f"'{date}'" if date else "NULL"
            # ON CONFLICT makes it idempotent: re-running won't create duplicates
            statements.append(
                f"INSERT INTO users (id, email, signup_date) "
                f"VALUES ({int(row['id'])}, '{email_sql}', {date_sql}) "
                f"ON CONFLICT (id) DO NOTHING;"
            )
    return statements
```

Maoni yanaonyesha sehemu muhimu: **uthibitisho** (ruka barua pepe tupu), **escaping** (manukuu) na **idempotence** (`ON CONFLICT DO NOTHING`). Omba AI kuingiza zote tatu — hii ndio vitu ambavyo script naïve inasahau.

## Kwa nini ni muhimu

Uhamiaji wa data una hatari ya juu na kawaida ni mara moja: script inayoingiza maiga au kuangusha safu inaweza kuwa ghali kutenganisha. AI inaharakisha kuandika ubadilisho, lakini mali ya usalama — idempotence, uthibitisho, na dry-run kwenye nakala — haziwezi kukubaliwa. Tilimu script iliyojifanya kama rasimu ambayo lazima useme na ujaribu, kamwe sio kitu unachoeneza kwa njia ya ushindi dhidi ya data halisi.