tests: added stats --cache-threshold negative arg tests

tests/test_stats.rs

@@ -896,7 +896,7 @@ fn stats_cache() {
  cmd.arg("--infer-dates")
  .arg("--dates-whitelist")
  .arg("all")
- // set cache threshold to 1 byte to force cache creation
+ // set cache threshold to 1 to force cache creation
  .args(["--cache-threshold", "1"])
  .arg(test_file);
 
@@ -918,6 +918,81 @@ fn stats_cache() {
  assert!(Path::new(&wrk.path("boston311-100.stats.csv.json")).exists());
 }
 
+#[test]
+fn stats_cache_negative_threshold() {
+ use std::path::Path;
+
+ let wrk = Workdir::new("stats_cache_negative_threshold");
+ let test_file = wrk.load_test_file("boston311-100.csv");
+
+ let mut cmd = wrk.command("stats");
+ cmd.arg("--infer-dates")
+ .arg("--dates-whitelist")
+ .arg("all")
+ // set cache threshold to -10240 to set autoindex_size to 10 kb
+ // and to force cache creation
+ .args(["-c", "-10240"])
+ .arg(test_file.clone());
+
+ let got: Vec<Vec<String>> = wrk.read_stdout(&mut cmd);
+
+ // the index file SHOULD have been created as the input file size > 10 kb
+ assert!(Path::new(&format!("{test_file}.idx")).exists());
+ // assert!(Path::new(&format!("{test_file}.idx")).exists());
+
+ wrk.create("in2.csv", got);
+
+ // removed variance & stddev columns as its causing flaky CI test for float values
+ let mut cmd = wrk.command("select");
+ cmd.arg("1-9,12-").arg("in2.csv");
+
+ let got2: String = wrk.stdout(&mut cmd);
+ let expected2 = wrk.load_test_resource("boston311-100-stats.csv");
+
+ assert_eq!(dos2unix(&got2), dos2unix(&expected2).trim_end());
+
+ // check that the stats cache files were created
+ assert!(Path::new(&wrk.path("boston311-100.stats.csv")).exists());
+ assert!(Path::new(&wrk.path("boston311-100.stats.csv.json")).exists());
+}
+
+#[test]
+fn stats_cache_negative_threshold_unmet() {
+ use std::path::Path;
+
+ let wrk = Workdir::new("stats_cache_negative_threshold_unmet");
+ let test_file = wrk.load_test_file("boston311-100.csv");
+
+ let mut cmd = wrk.command("stats");
+ cmd.arg("--infer-dates")
+ .arg("--dates-whitelist")
+ .arg("all")
+ // set cache threshold to -51200 to set autoindex_size to 50 kb
+ // and to force cache creation
+ .args(["--cache-threshold", "-51200"])
+ .arg(test_file.clone());
+
+ let got: Vec<Vec<String>> = wrk.read_stdout(&mut cmd);
+
+ // the index file SHOULD NOT have been created as the input file < 50 kb
+ assert!(!Path::new(&format!("{test_file}.idx")).exists());
+
+ wrk.create("in2.csv", got);
+
+ // removed variance & stddev columns as its causing flaky CI test for float values
+ let mut cmd = wrk.command("select");
+ cmd.arg("1-9,12-").arg("in2.csv");
+
+ let got2: String = wrk.stdout(&mut cmd);
+ let expected2 = wrk.load_test_resource("boston311-100-stats.csv");
+
+ assert_eq!(dos2unix(&got2), dos2unix(&expected2).trim_end());
+
+ // check that the stats cache files were created
+ assert!(Path::new(&wrk.path("boston311-100.stats.csv")).exists());
+ assert!(Path::new(&wrk.path("boston311-100.stats.csv.json")).exists());
+}
+
 #[test]
 fn stats_infer_boolean_1_0() {
  let wrk = Workdir::new("stats_infer_boolean_1_0");