SPARQL: adds support of REGEX

lib/Cargo.toml

@@ -28,6 +28,7 @@ rust_decimal = "0.10"
 chrono = "0.4"
 language-tags = "0.2"
 failure = "0.1"
+regex = "1"
 
 [build-dependencies]
 peg = "0.5"

lib/src/lib.rs

@@ -32,20 +32,21 @@
 )]
 
 extern crate byteorder;
-#[macro_use]
-extern crate lazy_static;
 extern crate chrono;
+#[macro_use]
+extern crate failure;
 extern crate language_tags;
+#[macro_use]
+extern crate lazy_static;
 extern crate num_traits;
 extern crate ordered_float;
 extern crate quick_xml;
+extern crate regex;
 #[cfg(feature = "rocksdb")]
 extern crate rocksdb;
 extern crate rust_decimal;
 extern crate url;
 extern crate uuid;
-#[macro_use]
-extern crate failure;
 
 pub mod model;
 pub mod rio;

lib/src/sparql/eval.rs

@@ -8,6 +8,7 @@ use num_traits::FromPrimitive;
 use num_traits::One;
 use num_traits::ToPrimitive;
 use ordered_float::OrderedFloat;
+use regex::RegexBuilder;
 use rust_decimal::Decimal;
 use sparql::algebra::*;
 use sparql::plan::*;
@@ -22,6 +23,8 @@ use store::encoded::EncodedQuadsStore;
 use store::numeric_encoder::*;
 use Result;
 
+const REGEX_SIZE_LIMIT: usize = 1_000_000;
+
 type EncodedTuplesIterator<'a> = Box<dyn Iterator<Item = Result<EncodedTuple>> + 'a>;
 
 pub struct SimpleEvaluator<S: EncodedQuadsStore> {
@@ -484,7 +487,36 @@ impl<S: EncodedQuadsStore> SimpleEvaluator<S> {
                     }.into(),
                 )
             }
-            PlanExpression::Regex(text, pattern, flags) => unimplemented!(),
+            PlanExpression::Regex(text, pattern, flags) => {
+                // TODO Avoid to compile the regex each time
+                let pattern = self.to_simple_string(self.eval_expression(pattern, tuple)?)?;
+                let mut regex_builder = RegexBuilder::new(&pattern);
+                regex_builder.size_limit(REGEX_SIZE_LIMIT);
+                if let Some(flags) = flags {
+                    let flags = self.to_simple_string(self.eval_expression(flags, tuple)?)?;
+                    for flag in flags.chars() {
+                        match flag {
+                            's' => {
+                                regex_builder.dot_matches_new_line(true);
+                            }
+                            'm' => {
+                                regex_builder.multi_line(true);
+                            }
+                            'i' => {
+                                regex_builder.case_insensitive(true);
+                            }
+                            'x' => {
+                                regex_builder.ignore_whitespace(true);
+                            }
+                            'q' => (), //TODO: implement
+                            _ => (),
+                        }
+                    }
+                }
+                let regex = regex_builder.build().ok()?;
+                let text = self.to_string(self.eval_expression(text, tuple)?)?;
+                Some(regex_builder.build().ok()?.is_match(&text).into())
+            }
             PlanExpression::BooleanCast(e) => match self.eval_expression(e, tuple)? {
                 EncodedTerm::BooleanLiteral(value) => Some(value.into()),
                 EncodedTerm::SimpleLiteral { value_id }
@@ -652,6 +684,19 @@ impl<S: EncodedQuadsStore> SimpleEvaluator<S> {
         }
     }
 
+    fn to_string(&self, term: EncodedTerm) -> Option<String> {
+        match term {
+            EncodedTerm::SimpleLiteral { value_id }
+            | EncodedTerm::StringLiteral { value_id }
+            | EncodedTerm::LangStringLiteral { value_id, .. } => Some(
+                str::from_utf8(&self.store.get_bytes(value_id).ok()??)
+                    .ok()?
+                    .to_owned(),
+            ),
+            _ => None,
+        }
+    }
+
     fn parse_numeric_operands(
         &self,
         e1: &PlanExpression,

lib/tests/sparql_test_cases.rs

@@ -82,7 +82,7 @@ fn sparql_w3c_syntax_testsuite() {
 
 #[test]
 fn sparql_w3c_query_evaluation_testsuite() {
-    //TODO: dataset graph open-world regex
+    //TODO: dataset graph open-world
     let manifest_10_urls = vec![
         Url::parse("http://www.w3.org/2001/sw/DataAccess/tests/data-r2/algebra/manifest.ttl")
             .unwrap(),
@@ -118,6 +118,8 @@ fn sparql_w3c_query_evaluation_testsuite() {
             .unwrap(),
         Url::parse("http://www.w3.org/2001/sw/DataAccess/tests/data-r2/reduced/manifest.ttl")
             .unwrap(),
+        Url::parse("http://www.w3.org/2001/sw/DataAccess/tests/data-r2/regex/manifest.ttl")
+            .unwrap(),
         Url::parse("http://www.w3.org/2001/sw/DataAccess/tests/data-r2/solution-seq/manifest.ttl")
             .unwrap(),
         Url::parse("http://www.w3.org/2001/sw/DataAccess/tests/data-r2/sort/manifest.ttl").unwrap(),