improve table create

This commit is contained in:
Dita Aji Pratama 2026-07-02 14:33:36 +07:00
parent 08289c4d43
commit 748e48bdf5

View File

@ -1,55 +1,11 @@
import embedding import lancedb, pyarrow
import config
import lancedb db = lancedb.connect(config.memories_db_path)
schema = pyarrow.schema([
url = "http://localhost:11434/api/embed" pyarrow.field( 'id', pyarrow.string() ),
model = "nomic-embed-text" pyarrow.field( 'title', pyarrow.string() ),
text = "Saya lupa password di HRIS AFMS2" pyarrow.field( 'story', pyarrow.string() ),
pyarrow.field( 'vector_title', pyarrow.list_( pyarrow.float32(), config.memories_vector_size ) ),
memories_db_path = "./memories" ])
memories_table = "knowledge_stories" db.create_table(config.memories_table, schema=schema)
print(f'Table "{config.memories_table}" berhasil diproses.')
docs = [
{
"id": "wifi-001",
"title": "Perubahan password WiFi menjadi login portal",
"story": (
"Password WiFi kantor telah berubah. "
"Sekarang akses WiFi menggunakan login portal. "
"Akun login akan diberikan melalui chat pribadi masing-masing user."
),
},
{
"id": "m365-001",
"title": "Cara cek spam pada Outlook Group",
"story": (
"Untuk mengecek spam pada Outlook Group, buka Outlook Web. "
"Masuk ke menu Groups, pilih group terkait, lalu cek folder Junk Email atau Spam."
),
},
{
"id": "printer-001",
"title": "Printer tidak terdeteksi di komputer",
"story": (
"Jika printer tidak terdeteksi, cek koneksi kabel atau jaringan. "
"Pastikan driver printer sudah terinstall, lalu coba restart service Print Spooler."
),
},
]
rows = []
for doc in docs:
row = {
"id": doc["id"],
"title": doc["title"],
"story": doc["story"],
"vector_title": embedding.embed_text(doc["title"]),
}
rows.append(row)
db = lancedb.connect(memories_db_path)
table = db.create_table(memories_table, data=rows, mode="overwrite")
print("Table berhasil dibuat.")
print("Nama table:", memories_table)
print("Jumlah row:", table.count_rows())