|
| 1 | +from pgml import Database |
| 2 | +import os |
| 3 | +from datasets import load_dataset |
| 4 | +from time import time |
| 5 | +from dotenv import load_dotenv |
| 6 | +from rich.console import Console |
| 7 | +from pypdf import PdfReader |
| 8 | +from pgml.dbutils import run_select_statement |
| 9 | + |
| 10 | +load_dotenv() |
| 11 | +console = Console() |
| 12 | + |
| 13 | +local_pgml = "postgres://postgres@127.0.0.1:5433/pgml_development" |
| 14 | + |
| 15 | +conninfo = os.environ.get("PGML_CONNECTION", local_pgml) |
| 16 | +db = Database(conninfo) |
| 17 | + |
| 18 | +collection_name = "pdf_collection" |
| 19 | +collection = db.create_or_get_collection(collection_name) |
| 20 | + |
| 21 | +filename = "lincoln.pdf" |
| 22 | +reader = PdfReader("lincoln.pdf") |
| 23 | +number_of_pages = len(reader.pages) |
| 24 | +documents = [] |
| 25 | +for page_number, page in enumerate(reader.pages): |
| 26 | + documents.append({"text": page.extract_text(),"page": page_number, "source": filename}) |
| 27 | + |
| 28 | +collection.upsert_documents(documents) |
| 29 | +collection.generate_chunks() |
| 30 | +collection.generate_embeddings() |
| 31 | + |
| 32 | +start = time() |
| 33 | +query = "When was Lincoln born?" |
| 34 | +results = collection.vector_search(query, top_k=1) |
| 35 | + |
| 36 | +# db.archive_collection(collection_name) |
| 37 | +conn = db.pool.getconn() |
| 38 | +context = " ".join(results[0]["chunk"].strip().split()) |
| 39 | +context = context.replace('"', '\\"').replace("'", "''") |
| 40 | + |
| 41 | +select_statement = """SELECT pgml.transform( |
| 42 | + 'question-answering', |
| 43 | + inputs => ARRAY[ |
| 44 | + '{ |
| 45 | + \"question\": \"%s\", |
| 46 | + \"context\": \"%s\" |
| 47 | + }' |
| 48 | + ] |
| 49 | +) AS answer;""" % ( |
| 50 | + query, |
| 51 | + context, |
| 52 | +) |
| 53 | + |
| 54 | +results = run_select_statement(conn, select_statement) |
| 55 | +db.pool.putconn(conn) |
| 56 | + |
| 57 | +console.print("\nResults for query '%s'" % query) |
| 58 | +console.print(results) |
| 59 | +db.archive_collection(collection_name) |
| 60 | +_end = time() |
| 61 | +console.print("Query time = %0.3f" % (_end - start)) |
0 commit comments