Start adding row-level splitting, refactor cli and graph into subcrates

This commit is contained in:
2024-08-09 22:13:43 +09:30
parent 3cdaa81da1
commit 0ee88e3a99
11 changed files with 259 additions and 110 deletions

82
src/graph/sql_rule.rs Normal file
View File

@@ -0,0 +1,82 @@
use std::fs::File;
use polars::{
io::SerWriter,
prelude::{CsvWriter, LazyCsvReader, LazyFileListReader},
};
use polars_sql::SQLContext;
use schemars::JsonSchema;
use serde::{Deserialize, Serialize};
use super::node::RunnableNode;
#[derive(Serialize, Deserialize, Clone, JsonSchema)]
pub struct CSVFile {
name: String,
path: String,
}
/**
* Run SQL over files using polars, export results to output file
*/
fn run_sql(files: &Vec<CSVFile>, output_path: &String, query: &String) -> anyhow::Result<()> {
let mut ctx = SQLContext::new();
for file in files {
let df = LazyCsvReader::new(&file.path).finish()?;
ctx.register(&file.name, df);
}
let result = ctx.execute(&query)?;
let mut file = File::create(output_path)?;
CsvWriter::new(&mut file).finish(&mut result.collect()?)?;
Ok(())
}
#[derive(Serialize, Deserialize, Clone, JsonSchema)]
pub struct SQLNode {
pub files: Vec<CSVFile>,
pub output_file: String,
pub query: String,
}
pub struct SQLNodeRunner {
pub sql_node: SQLNode,
}
impl RunnableNode for SQLNodeRunner {
fn run(&self) -> anyhow::Result<()> {
run_sql(
&self.sql_node.files,
&self.sql_node.output_file,
&self.sql_node.query,
)
}
}
#[cfg(test)]
mod tests {
use std::{fs::File, io::Read};
use super::{run_sql, CSVFile};
#[test]
fn basic_query_works() -> anyhow::Result<()> {
let output_path = "./testing/output/output.csv".to_owned();
run_sql(
&vec![CSVFile {
name: "Account".to_owned(),
path: "./testing/test.csv".to_owned(),
}],
&output_path,
&"SELECT * FROM Account WHERE Code = 'A195950'".to_owned(),
)?;
let mut output = String::new();
let mut output_file = File::open(output_path)?;
output_file.read_to_string(&mut output)?;
assert_eq!(
output,
"Code,Description,Type,CostOutput,PercentFixed
A195950,A195950 Staff Related Other,E,GS,100.00
"
);
Ok(())
}
}