From: Erik Mackdanz <erikmack@gmail.com>
Date: Sat, 25 Nov 2023 03:40:15 +0000 (-0600)
Subject: three partitions not two
X-Git-Url: https://git.humopery.space/?a=commitdiff_plain;h=5d4e990db3cb8aeaacefa0b236a98d8b9425594a;p=polarsbop.git

three partitions not two
---

diff --git a/scripts/paraexport b/scripts/paraexport
index 972e7cf..be33a5d 100755
--- a/scripts/paraexport
+++ b/scripts/paraexport
@@ -3,7 +3,8 @@
 # start multiple jobs in parallel
 target/debug/polarsbop 0 &
 target/debug/polarsbop 1 &
+target/debug/polarsbop 2 &
 
-wait %1 %2
+wait %1 %2 %3
 
 echo done
diff --git a/src/main.rs b/src/main.rs
index 0182cc6..eb31d5e 100644
--- a/src/main.rs
+++ b/src/main.rs
@@ -4,16 +4,16 @@ use std::fs::{create_dir_all,File};
 
 fn main() -> Result<(),PolarsError> {
 
-    // expect first arg to be a file suffix = 0, 1
+    // expect first arg to be a file partition = 0, 1, 2
     let mut argsiter = args();
     let _ = argsiter.next();
-    let suffix = argsiter.next()
-	.expect("single argument required, 0 or 1");
+    let partition = argsiter.next()
+	.expect("single argument required, 0 or 1 or 2");
 
     // prep output file
     create_dir_all("out")
 	.expect("couldn't create out dir");
-    let outfilename = format!("out/output-{}.json",suffix);
+    let outfilename = format!("out/output-{}.json",partition);
     // println!("{}",outfilename);
     let mut file = File::create(outfilename)
 	.expect("couldn't create output file");
@@ -38,18 +38,30 @@ fn main() -> Result<(),PolarsError> {
 	)));
 
     // read and parse from disk
-    let df = CsvReader::from_path("data/bop.csv")?
+    let mut df = CsvReader::from_path("data/bop.csv")?
 	.has_header(true)
 	.with_dtypes(schema_override)
 	.with_null_values(null_values)
 	.finish()?;
 
     // split list based on arg
-    let mut mask = df.column("Country Name")?.lt("Luxemb")?;
-    if suffix == "1" {
-	mask = df.column("Country Name")?.gt_eq("Luxemb")?;
+    match partition.as_str() {
+	"0" => {
+	    let mask = df.column("Country Name")?.lt("Ethiopia")?;
+	    df = df.filter(&mask)?;
+	},
+	"1" => {
+	    let mask = df.column("Country Name")?.gt_eq("Ethiopia")?;
+	    df = df.filter(&mask)?;
+	    let mask = df.column("Country Name")?.lt("Romania")?;
+	    df = df.filter(&mask)?;
+	},
+	"2" => {
+	    let mask = df.column("Country Name")?.gt_eq("Romania")?;
+	    df = df.filter(&mask)?;
+	},
+	_ => { panic!("partition must be 0, 1 or 2"); }
     }
-    let df = df.filter(&mask)?;
 
     // values not status, selected cols
     let mut df = df