Add built-in functions

Cargo.toml

@@ -62,8 +62,9 @@ strum = "0.28.0"
 strum_macros = "0.28.0"
 test-log = "0.2"
 thiserror = "2.0"
-tokio = "1.47.1"
+tokio = { version = "1.47.1", features = ["rt", "macros"] }
 urlencoding = "2.1.3"
+uuid = { version = "1", features = ["v7"] }
 line-index = "0.1.1"
 
 [profile.dev]

nemo-physical/Cargo.toml

@@ -12,6 +12,7 @@ repository.workspace = true
 [features]
 check_column_sorting = []
 old_dictionaries = []
+js = ["uuid/js", "oxsdatatypes/js"]
 
 [[bin]]
 name = "dict-bench"
@@ -31,7 +32,6 @@ lru = "0.16"
 cpu-time = "1.0"
 reqwest = { workspace = true }
 delegate = { workspace = true }
-regex = "1.9.5"
 hashbrown = "0.16.0"
 streaming-iterator = "0.1.9"
 flate2 = { workspace = true }
@@ -40,9 +40,17 @@ urlencoding = { workspace = true }
 oxiri = { workspace = true }
 path-slash = "0.2.1"
 levenshtein = "1.0.5"
+rand = { workspace = true }
+uuid = { workspace = true }
+sha2 = "0.10"
+sha1 = "0.10"
+md5 = { package = "md-5", version = "0.10" }
+digest = "0.10"
 serde_urlencoded = "0.7.1"
 serde = { workspace = true, features =["derive"]}
 itertools = { workspace = true }
+oxsdatatypes = "0.2.2"
+fancy-regex = "0.18"
 
 [dev-dependencies]
 rand = { workspace = true }

nemo-physical/src/datatypes/storage_type_name.rs

@@ -24,7 +24,7 @@ pub(crate) enum StorageTypeName {
 
 /// A list of [StorageTypeName],
 /// in the order they appear in the enum.
-pub(crate) const STORAFE_TYPES: &[StorageTypeName] = &[
+pub(crate) const STORAGE_TYPES: &[StorageTypeName] = &[
     StorageTypeName::Id32,
     StorageTypeName::Id64,
     StorageTypeName::Int64,
@@ -113,6 +113,14 @@ impl StorageTypeBitSet {
         result
     }
 
+    /// Return the first [StorageTypeName] contained in this set, if any.
+    pub(crate) fn first_type(&self) -> Option<StorageTypeName> {
+        STORAGE_TYPES
+            .iter()
+            .copied()
+            .find(|storage_type| self.contains(storage_type))
+    }
+
     /// Returns `true` if there is exactly one possible type.
     pub(crate) fn is_unique(&self) -> bool {
         self.0.is_single()

nemo-physical/src/datavalues/any_datavalue.rs

@@ -615,6 +615,8 @@ impl DataValue for AnyDataValue {
             fn to_boolean_unchecked(&self) -> bool;
             fn to_null(&self) -> Option<NullDataValue>;
             fn to_null_unchecked(&self) -> NullDataValue;
+            fn to_other(&self) -> Option<(String, String)>;
+            fn to_other_unchecked(&self) -> (String, String);
             fn label(&self) -> Option<&IriDataValue>;
             fn length(&self) -> Option<usize>;
             fn len_unchecked(&self) -> usize;

nemo-physical/src/datavalues/datavalue.rs

@@ -487,6 +487,24 @@ pub trait DataValue: Debug + Display + Into<AnyDataValue> + PartialEq + Eq + Has
         panic!("Value is not a null.");
     }
 
+    /// If this value is an other literal, return the pair of its lexical value and the IRI of its datatype.
+    #[must_use]
+    fn to_other(&self) -> Option<(String, String)> {
+        match self.value_domain() {
+            ValueDomain::Other => Some(self.to_other_unchecked()),
+            _ => None,
+        }
+    }
+
+    /// If this value is an other literal, return the pair of its lexical value and the IRI of its datatype.
+    ///
+    /// # Panics
+    /// Panics if this value is not an other literal.
+    #[must_use]
+    fn to_other_unchecked(&self) -> (String, String) {
+        panic!("Value is not an other literal.");
+    }
+
     /// Return the IRI-valued label of this complex value if it has a label,
     /// and None otherwise. Only values in the domain [ValueDomain::Tuple] or
     /// [ValueDomain::Map] can have labels.

nemo-physical/src/datavalues/other_datavalue.rs

@@ -34,6 +34,10 @@ impl DataValue for OtherDataValue {
             + RDF_DATATYPE_INDICATOR
             + &super::datavalue::quote_iri(self.1.as_str())
     }
+
+    fn to_other_unchecked(&self) -> (String, String) {
+        (self.0.to_owned(), self.1.to_owned())
+    }
 }
 
 impl std::hash::Hash for OtherDataValue {