Initial commit

crates/smol-uca/tests/collation_test.rs

@@ -0,0 +1,91 @@
+use std::fs::File;
+use std::io::{BufRead, BufReader};
+
+use smol_uca::collator::Collator;
+
+#[test]
+fn collation_test_non_ignorable() {
+    let data = File::open("data/CollationTest_NON_IGNORABLE.txt")
+        .map(BufReader::new)
+        .expect("collation test data");
+
+    let collator = Collator::default();
+
+    let mut prev_sort_key = None;
+
+    let mut order_errors = 0;
+    let mut sort_key_errors = 0;
+
+    'line: for (n, line) in data.lines().enumerate() {
+        let line = line.expect("line");
+
+        let line = line.trim_start();
+
+        if line.is_empty() || line.starts_with('#') {
+            continue;
+        }
+
+        let (chars, rest) = line
+            .split_once(';')
+            .expect("a semicolon separated test line");
+
+        let mut surrogates = false;
+        let test_string = chars
+            .trim()
+            .split(' ')
+            .map(|x| u32::from_str_radix(x, 16).expect("a valid hex value"))
+            .map(|x| match char::from_u32(x) {
+                Some(ch) => ch,
+                None => {
+                    if (0xD800u32..=0xDFFF).contains(&x) {
+                        surrogates = true;
+
+                        ' '
+                    } else {
+                        panic!("{}", line)
+                    }
+                }
+            })
+            .collect::<String>();
+
+        if surrogates {
+            continue 'line;
+        }
+
+        let expected_sort_key = rest.rsplit(['[', ']']).nth(1).expect("sort key");
+
+        let sort_key = collator.sort_key(&test_string);
+        let fmt_sort_key = smol_uca::fmt(&sort_key);
+
+        if let Some(prev_sort_key) = prev_sort_key.take() {
+            if sort_key < prev_sort_key {
+                eprintln!(
+                    "Error at line {}: {:?} [{}]",
+                    n + 1,
+                    test_string,
+                    expected_sort_key
+                );
+
+                order_errors += 1;
+            }
+        }
+
+        prev_sort_key = Some(sort_key);
+
+        if fmt_sort_key != expected_sort_key {
+            eprintln!(
+                "Error at line {}: {:?} expected: [{}], got: [{}] ({})",
+                n + 1,
+                unf::nfd(&test_string).collect::<String>(),
+                expected_sort_key,
+                fmt_sort_key,
+                line
+            );
+
+            sort_key_errors += 1;
+        }
+    }
+
+    assert_eq!(order_errors, 0);
+    assert_eq!(sort_key_errors, 0);
+}