split: add --no-headers logic for --kb-size option

src/cmd/split.rs

@@ -166,10 +166,14 @@ impl Args {
  let mut headerbuf_wtr = csv::WriterBuilder::new().from_writer(vec![]);
 
  headerbuf_wtr.write_byte_record(&headers)?;
- // safety: we know the inner vec is valid
- let header_string =
- simdutf8::basic::from_utf8(&headerbuf_wtr.into_inner().unwrap())?.to_string();
- let header_byte_size = header_string.len();
+ let header_byte_size = if self.flag_no_headers {
+ 0
+ } else {
+ // safety: we know the inner vec is valid
+ let header_string =
+ simdutf8::basic::from_utf8(&headerbuf_wtr.into_inner().unwrap())?.to_string();
+ header_string.len()
+ };
 
  let mut wtr = self.new_writer(&headers, 0, self.flag_pad)?;
  let mut i = 0;

tests/test_split.rs

@@ -780,8 +780,8 @@ fn split_nooutdir() {
 }
 
 #[test]
-fn split_kbsize_boston_2k() {
- let wrk = Workdir::new("split_kbsize_boston_2k");
+fn split_kbsize_boston_5k() {
+ let wrk = Workdir::new("split_kbsize_boston_5k");
  let test_file = wrk.load_test_file("boston311-100.csv");
 
  let mut cmd = wrk.command("split");
@@ -805,8 +805,8 @@ fn split_kbsize_boston_2k() {
 }
 
 #[test]
-fn split_kbsize_boston_2k_padded() {
- let wrk = Workdir::new("split_kbsize_boston_2k_padded");
+fn split_kbsize_boston_5k_padded() {
+ let wrk = Workdir::new("split_kbsize_boston_5k_padded");
  let test_file = wrk.load_test_file("boston311-100.csv");
 
  let mut cmd = wrk.command("split");
@@ -830,3 +830,28 @@ fn split_kbsize_boston_2k_padded() {
  assert!(wrk.path("testme-086.csv").exists());
  assert!(wrk.path("testme-095.csv").exists());
 }
+
+#[test]
+fn split_kbsize_boston_5k_no_headers() {
+ let wrk = Workdir::new("split_kbsize_boston_5k_no_headers");
+ let test_file = wrk.load_test_file("boston311-100.csv");
+
+ let mut cmd = wrk.command("split");
+ cmd.args(["--kb-size", "5"])
+ .arg(&wrk.path("."))
+ .arg("--no-headers")
+ .arg(test_file);
+ wrk.run(&mut cmd);
+
+ assert!(wrk.path("0.csv").exists());
+ assert!(wrk.path("12.csv").exists());
+ assert!(wrk.path("21.csv").exists());
+ assert!(wrk.path("29.csv").exists());
+ assert!(wrk.path("39.csv").exists());
+ assert!(wrk.path("48.csv").exists());
+ assert!(wrk.path("56.csv").exists());
+ assert!(wrk.path("66.csv").exists());
+ assert!(wrk.path("76.csv").exists());
+ assert!(wrk.path("84.csv").exists());
+ assert!(wrk.path("93.csv").exists());
+}