implement SPARQL functions UCASE and LCASE

pchampin · Dec 9, 2024 · 501f0df · 501f0df
1 parent 2a9f0e8
commit 501f0df
Show file tree

Hide file tree

Showing 3 changed files with 71 additions and 3 deletions.
diff --git a/sparql/src/function.rs b/sparql/src/function.rs
@@ -117,8 +117,18 @@ pub fn call_function(function: &Function, mut arguments: Vec<EvalResult>) -> Opt
             Some(str_len(string.as_string_lit()?.0))
         }
         Replace => todo("Replace"),
-        UCase => todo("UCase"),
-        LCase => todo("LCase"),
+        UCase => {
+            let [string] = &arguments[..] else {
+                unreachable!();
+            };
+            Some(u_case(string.as_string_lit()?))
+        }
+        LCase => {
+            let [string] = &arguments[..] else {
+                unreachable!();
+            };
+            Some(l_case(string.as_string_lit()?))
+        }
         EncodeForUri => todo("EncodeForUri"),
         Contains => todo("Contains"),
         StrStarts => todo("StrStarts"),
@@ -329,6 +339,16 @@ pub fn str_len(string: &Arc<str>) -> EvalResult {
     }
 }
 
+pub fn u_case(source: (&Arc<str>, Option<&LanguageTag<Arc<str>>>)) -> EvalResult {
+    let lex: String = source.0.chars().flat_map(char::to_uppercase).collect();
+    EvalResult::from((Arc::from(lex), source.1.cloned()))
+}
+
+pub fn l_case(source: (&Arc<str>, Option<&LanguageTag<Arc<str>>>)) -> EvalResult {
+    let lex: String = source.0.chars().flat_map(char::to_lowercase).collect();
+    EvalResult::from((Arc::from(lex), source.1.cloned()))
+}
+
 pub fn triple(s: &EvalResult, p: &EvalResult, o: &EvalResult) -> Option<EvalResult> {
     let EvalResult::Term(s) = s else { return None };
     let EvalResult::Term(p) = p else { return None };

diff --git a/sparql/src/function/test.rs b/sparql/src/function/test.rs
@@ -193,12 +193,45 @@ fn sub_str(source: &str, start: f64, length: Option<f64>, exp: Option<&str>) ->
 fn str_len(string: &str, exp: isize) -> TestResult {
     let pair = txt2pair(string);
     let string = &pair.0;
-    let source = (&pair.0, pair.1.as_ref());
     let exp = EvalResult::from(SparqlNumber::from(exp));
     assert!(eval_eq(Some(super::str_len(string)), Some(exp)));
     Ok(())
 }
 
+#[test_case("foo", "FOO")]
+#[test_case("foo@en", "FOO@en")]
+#[test_case("FOO", "FOO"; "noop")]
+#[test_case("FOO@en", "FOO@en"; "noop en")]
+#[test_case("fooBAR 1!xY", "FOOBAR 1!XY")]
+#[test_case("fooBAR 1!xY@en", "FOOBAR 1!XY@en")]
+#[test_case("àéîôù", "ÀÉÎÔÙ"; "accents")]
+#[test_case("àéîôù@fr", "ÀÉÎÔÙ@fr"; "accents fr")]
+#[test_case("ﬀ ŉ", "FF ʼN"; "multichar")]
+#[test_case("ﬀ ŉ@en", "FF ʼN@en"; "multichar en")]
+fn u_case(string: &str, exp: &str) -> TestResult {
+    let pair = txt2pair(string);
+    let source = (&pair.0, pair.1.as_ref());
+    let exp = EvalResult::from(txt2pair(exp));
+    assert!(eval_eq(Some(super::u_case(source)), Some(exp)));
+    Ok(())
+}
+
+#[test_case("FOO", "foo")]
+#[test_case("FOO@en", "foo@en")]
+#[test_case("foo", "foo"; "noop")]
+#[test_case("foo@en", "foo@en"; "noop en")]
+#[test_case("fooBAR 1!xY", "foobar 1!xy")]
+#[test_case("fooBAR 1!xY@en", "foobar 1!xy@en")]
+#[test_case("ÀÉÎÔÙ", "àéîôù"; "accents")]
+#[test_case("ÀÉÎÔÙ@fr", "àéîôù@fr"; "accents fr")]
+fn l_case(string: &str, exp: &str) -> TestResult {
+    let pair = txt2pair(string);
+    let source = (&pair.0, pair.1.as_ref());
+    let exp = EvalResult::from(txt2pair(exp));
+    assert!(eval_eq(Some(super::l_case(source)), Some(exp)));
+    Ok(())
+}
+
 #[test_case("<tag:s>", "<tag:p>", "<tag:o>", true)]
 #[test_case("<tag:s>", "<tag:p>", "bnode()", true)]
 #[test_case("<tag:s>", "<tag:p>", " \"o\" ", true)]

diff --git a/sparql/src/test.rs b/sparql/src/test.rs
@@ -495,6 +495,21 @@ fn test_expr_variable() -> TestResult {
 #[test_case("strLen(\"foobar\"@en)", "6"; "strLen for language string")]
 #[test_case("strLen(42)", ""; "strLen for number")]
 #[test_case("strLen(<< <tag:s> <tag:p> <tag:o> >>)", ""; "strLen for triple")]
+// TODO test replace
+// test uCase
+#[test_case("uCase(<tag:x>)", ""; "uCase for IRI")]
+#[test_case("uCase(bnode())", ""; "uCase for bnode")]
+#[test_case("uCase(\"fooBAR\")", "\"FOOBAR\""; "uCase for string")]
+#[test_case("uCase(\"fooBAR\"@en)", "\"FOOBAR\"@en"; "uCase for language string")]
+#[test_case("uCase(42)", ""; "uCase for number")]
+#[test_case("uCase(<< <tag:s> <tag:p> <tag:o> >>)", ""; "uCase for triple")]
+// test lCase
+#[test_case("lCase(<tag:x>)", ""; "lCase for IRI")]
+#[test_case("lCase(bnode())", ""; "lCase for bnode")]
+#[test_case("lCase(\"fooBAR\")", "\"foobar\""; "lCase for string")]
+#[test_case("lCase(\"fooBAR\"@en)", "\"foobar\"@en"; "lCase for language string")]
+#[test_case("lCase(42)", ""; "lCase for number")]
+#[test_case("lCase(<< <tag:s> <tag:p> <tag:o> >>)", ""; "lCase for triple")]
 // TODO test other function calls
 // test isIri
 #[test_case("isIri(<tag:x>)", "true"; "isIri for IRI")]