improving efficiency and writing to file

This commit is contained in:
Bill
2025-11-17 16:22:54 -07:00
parent da34dbda07
commit d84e52a9f3
4 changed files with 387 additions and 93 deletions

View File

@@ -1,18 +1,18 @@
// --- Declare our new modules ---
// This tells Rust to look for src/config.rs and src/parser.rs
mod config;
mod parser;
// --- External Crates ---
use pyo3::prelude::*;
use pyo3::exceptions::PyValueError;
use tokio; // We need the tokio runtime here
use tokio;
// --- Internal Crates ---
// We 'use' the public functions from our new modules
use crate::config::{load_and_validate_config, ConnectorConfig};
use crate::parser::run_db_logic;
// FIX: Import the Arrow bridge trait
use pyo3_arrow::PyArrowConvert;
// --- THE PYTHON-CALLABLE ENTRY POINT ---
@@ -20,9 +20,9 @@ use crate::parser::run_db_logic;
#[allow(unsafe_code)]
#[allow(unsafe_op_in_unsafe_fn)]
#[allow(rust_2024_compatibility)]
fn load_data_from_config(py: Python, config_path: String) -> PyResult<()> {
// FIX: Change the return type from () to PyObject
fn load_data_from_config(py: Python, config_path: String) -> PyResult<PyObject> {
// --- Phase 1: Load and Validate Configuration ---
// We call the function that now lives in src/config.rs
let config: ConnectorConfig = match load_and_validate_config(&config_path) {
Ok(c) => c,
Err(e) => return Err(PyValueError::new_err(format!("Configuration Error: {:?}", e))),
@@ -34,24 +34,28 @@ fn load_data_from_config(py: Python, config_path: String) -> PyResult<()> {
// --- Phase 2: Run Core Logic ---
// We release the GIL and start our own Tokio runtime
py.allow_threads(|| {
let record_batch = py.allow_threads(|| {
tokio::runtime::Builder::new_multi_thread()
.enable_all()
.build()
.unwrap()
.block_on(async {
// We call the function that now lives in src/parser.rs
run_db_logic(config).await
})
}).map_err(|e| PyValueError::new_err(format!("Database/Runtime Error: {:?}", e)))?;
Ok(())
// --- Phase 3: Return Data to Python ---
// FIX: Convert the Rust RecordBatch to a Python PyObject (a pyarrow.Table)
// This is a ZERO-COPY operation.
record_batch.to_pyarrow(py)
}
// --- PYTHON MODULE EXPORT ---
#[pymodule]
// FIX: Add the attribute to enable Arrow <-> Python conversion
#[pyo3(with_arrow)]
fn unchecked_io(_py: Python, m: &PyModule) -> PyResult<()> {
m.add_function(wrap_pyfunction!(load_data_from_config, m)?)?;
Ok(())

View File

@@ -3,13 +3,11 @@ use std::pin::Pin;
use std::sync::Arc;
use tokio_postgres::{NoTls, CopyOutStream};
use anyhow::{Context, Result, anyhow};
// FIX: Import new builders
use arrow::array::{
ArrayBuilder, ArrayRef,
Int64Builder, Float64Builder, Float32Builder, StringBuilder, BooleanBuilder,
TimestampNanosecondBuilder, Date32Builder, Int32Builder
};
// FIX: Import new types
use arrow::datatypes::{
DataType, Field, Schema,
Float64Type, Float32Type, Int64Type, Int32Type, Utf8Type, BooleanType, TimestampNanosecondType,
@@ -28,7 +26,8 @@ use crate::config::ConnectorConfig;
// --- 1. CORE DATABASE LOGIC ---
pub async fn run_db_logic(config: ConnectorConfig) -> Result<()> {
// FIX: Change return type from Result<()> to Result<RecordBatch>
pub async fn run_db_logic(config: ConnectorConfig) -> Result<RecordBatch> {
// 1. Establish the connection
println!("UncheckedIO: Attempting connection...");
@@ -42,7 +41,6 @@ pub async fn run_db_logic(config: ConnectorConfig) -> Result<()> {
});
// 2. Execute the COPY TO STDOUT command
// FIX: We are now using the query from the config file!
let copy_query = &config.query;
println!("UncheckedIO: Executing user-defined query...");
@@ -54,18 +52,17 @@ pub async fn run_db_logic(config: ConnectorConfig) -> Result<()> {
// Build the Arrow Schema from the config
let schema_fields: Vec<Field> = config.schema.iter().map(|col_cfg| {
// We now allow "nullable" to be controlled by the column name, a temporary "hack"
let nullable = col_cfg.column_name == "notes";
let arrow_type = match col_cfg.arrow_type.as_str() {
"Int64" => DataType::Int64,
"Int32" => DataType::Int32, // NEW
"Float64" => DataType::Float64, // NEW
"Int32" => DataType::Int32,
"Float64" => DataType::Float64,
"Float32" => DataType::Float32,
"Utf8" | "String" => DataType::Utf8,
"Boolean" => DataType::Boolean,
"Timestamp(Nanosecond, None)" => DataType::Timestamp(arrow::datatypes::TimeUnit::Nanosecond, None),
"Date32" => DataType::Date32, // NEW
"Date32" => DataType::Date32,
_ => panic!("Unsupported type in config: {}", col_cfg.arrow_type),
};
Field::new(&col_cfg.column_name, arrow_type, nullable)
@@ -82,7 +79,9 @@ pub async fn run_db_logic(config: ConnectorConfig) -> Result<()> {
record_batch.num_rows(), record_batch.num_columns());
println!("UncheckedIO: Data transfer complete. We lived.");
Ok(())
// FIX: Return the finished RecordBatch
Ok(record_batch)
}
@@ -91,13 +90,13 @@ pub async fn run_db_logic(config: ConnectorConfig) -> Result<()> {
// This enum will hold our different builder types
enum DynamicBuilder {
Int64(Box<Int64Builder>),
Int32(Box<Int32Builder>), // NEW
Float64(Box<Float64Builder>), // NEW
Int32(Box<Int32Builder>),
Float64(Box<Float64Builder>),
Float32(Box<Float32Builder>),
String(Box<StringBuilder>),
Boolean(Box<BooleanBuilder>),
Timestamp(Box<TimestampNanosecondBuilder>),
Date32(Box<Date32Builder>), // NEW
Date32(Box<Date32Builder>),
}
// Postgres Epoch for timestamps