From 222539a86471c10eebfe753af4b43a1479f9f0ad Mon Sep 17 00:00:00 2001
From: Orion Yeung <11580988+orionyeung001@users.noreply.github.com>
Date: Mon, 24 Jun 2024 19:27:51 -0500
Subject: [PATCH] test: improve ergonomics for test

---
 .gitignore                |  3 +++
 tests/gather_nist_data.sh | 16 ++++++++++------
 tests/nist_tests.rs       | 30 ++++++++++++++++--------------
 3 files changed, 29 insertions(+), 20 deletions(-)

diff --git a/.gitignore b/.gitignore
index 3bb686bd..fe9e65ec 100644
--- a/.gitignore
+++ b/.gitignore
@@ -7,6 +7,9 @@
 # Executables
 *.exe
 
+# Test data for integration tests
+tests/*.dat
+
 # Generated by Cargo
 /target/
 *.lock
diff --git a/tests/gather_nist_data.sh b/tests/gather_nist_data.sh
index 9381ff08..2b663734 100755
--- a/tests/gather_nist_data.sh
+++ b/tests/gather_nist_data.sh
@@ -1,27 +1,31 @@
 #! /bin/bash
+# this script is to download and preprocess datafiles for the nist_tests.rs
+# integration test for statrs downloads data to directory specified by env
+# var STATRS_NIST_DATA_DIR
 
 process_file() {
   # Define input and output file names
   SOURCE=$1
   FILENAME=$2
-  curl -fsSL ${SOURCE}/$FILENAME > $FILENAME
+  TARGET=${STATRS_NIST_DATA_DIR-tests}/${FILENAME}
+  echo -e ${FILENAME} '\n\tDownloading...'
+  curl -fsSL ${SOURCE}/$FILENAME > ${TARGET}
 
   # Extract line numbers for Certified Values and Data from the header
-  INFO=$(grep "Certified Values:" $FILENAME)
+  INFO=$(grep "Certified Values:" $TARGET)
   CERTIFIED_VALUES_START=$(echo $INFO | awk '{print $4}')
   CERTIFIED_VALUES_END=$(echo $INFO | awk '{print $6}')
 
-  INFO=$(grep "Data            :" $FILENAME)
+  INFO=$(grep "Data            :" $TARGET)
   DATA_START=$(echo $INFO | awk '{print $4}')
   DATA_END=$(echo $INFO | awk '{print $6}')
 
+  echo -e '\tFormatting...'
   # Extract and reformat sections
-  # Certified values
   sed -n -i \
       -e "${CERTIFIED_VALUES_START},${CERTIFIED_VALUES_END}p" \
       -e "${DATA_START},${DATA_END}p" \
-      $FILENAME
-
+      $TARGET
 }
 
 URL='https://www.itl.nist.gov/div898/strd/univ/data'
diff --git a/tests/nist_tests.rs b/tests/nist_tests.rs
index 9efc94bf..a985820f 100644
--- a/tests/nist_tests.rs
+++ b/tests/nist_tests.rs
@@ -45,32 +45,34 @@ const FILENAMES: [&str; 7] = [
     "NumAcc3.dat",
 ];
 
+fn get_path(fname: &str, prefix: Option<&str>) -> PathBuf {
+    if let Some(prefix) = prefix {
+        [prefix, fname].iter().collect()
+    } else {
+        ["tests", fname].iter().collect()
+    }
+}
+
 #[test]
 #[ignore = "NIST tests should not run from typical `cargo test` calls"]
 fn nist_strd_univariate_mean() {
-    let path_prefix = env::var(NIST_DATA_DIR_ENV).unwrap_or_else(|e| panic!("{}", e));
-
     for fname in FILENAMES {
-        let case = parse_file([&path_prefix, fname].iter().collect::<PathBuf>())
-            .unwrap_or_else(|e| panic!("failed parsing file {} with {:?}", fname, e));
-        assert_relative_eq!(
-            case.values.iter().mean(),
-            case.certified.mean,
-            epsilon = 1e-12
-        );
+        let filepath = get_path(fname, env::var(NIST_DATA_DIR_ENV).ok().as_deref());
+        let case = parse_file(filepath)
+            .unwrap_or_else(|e| panic!("failed parsing file {} with `{:?}`", fname, e));
+        assert_relative_eq!(case.values.mean(), case.certified.mean, epsilon = 1e-12);
     }
 }
 
 #[test]
 #[ignore]
 fn nist_strd_univariate_std_dev() {
-    let path_prefix = env::var(NIST_DATA_DIR_ENV).unwrap_or_else(|e| panic!("{}", e));
-
     for fname in FILENAMES {
-        let case = parse_file([&path_prefix, fname].iter().collect::<PathBuf>())
-            .unwrap_or_else(|e| panic!("failed parsing file {} with {:?}", fname, e));
+        let filepath = get_path(fname, env::var(NIST_DATA_DIR_ENV).ok().as_deref());
+        let case = parse_file(filepath)
+            .unwrap_or_else(|e| panic!("failed parsing file {} with `{:?}`", fname, e));
         assert_relative_eq!(
-            case.values.iter().std_dev(),
+            case.values.std_dev(),
             case.certified.std_dev,
             epsilon = 1e-10
         );