feat: detect simple cases of empty intersections

Author: Aloso

pomsky-lib/src/diagnose/compile_error.rs

@@ -96,6 +96,7 @@ pub(crate) enum CompileErrorKind {
     },
     NestedTest,
     BadIntersection,
+    EmptyIntersection,
 }
 
 impl CompileErrorKind {
@@ -114,11 +115,6 @@ impl core::fmt::Display for CompileErrorKind {
     fn fmt(&self, f: &mut core::fmt::Formatter<'_>) -> core::fmt::Result {
         match self {
             CompileErrorKind::ParseError(kind) => write!(f, "Parse error: {kind}"),
-            CompileErrorKind::BadIntersection => write!(
-                f,
-                "Intersecting these expressions is not supported. Only character sets \
-                can be intersected."
-            ),
             CompileErrorKind::Unsupported(feature, flavor) => match feature {
                 Feature::SpecificUnicodeProp => write!(
                     f,
@@ -223,6 +219,14 @@ impl core::fmt::Display for CompileErrorKind {
                 ),
                 _ => write!(f, "This kind of lookbehind is not supported in the {flavor:?} flavor"),
             },
+            CompileErrorKind::BadIntersection => write!(
+                f,
+                "Intersecting these expressions is not supported. Only character sets \
+                can be intersected."
+            ),
+            CompileErrorKind::EmptyIntersection => {
+                write!(f, "Intersection of expressions that do not overlap")
+            }
         }
     }
 }

pomsky-lib/src/diagnose/diagnostic_code.rs

@@ -99,6 +99,7 @@ diagnostic_code! {
         UnsupportedInLookbehind = 320,
         LookbehindNotConstantLength = 321,
         BadIntersection = 322,
+        EmptyIntersection = 323,
 
         // Warning indicating something might not be supported
         PossiblyUnsupported = 400,
@@ -234,6 +235,7 @@ impl<'a> From<&'a CompileErrorKind> for DiagnosticCode {
             C::LookbehindNotConstantLength { .. } => Self::LookbehindNotConstantLength,
             C::NestedTest => Self::NestedTest,
             C::BadIntersection => Self::BadIntersection,
+            C::EmptyIntersection => Self::EmptyIntersection,
         }
     }
 }

pomsky-lib/src/diagnose/diagnostic_kind.rs

@@ -40,7 +40,9 @@ impl From<&CompileErrorKind> for DiagnosticKind {
             | K::NameUsedMultipleTimes(_)
             | K::UnknownVariable { .. }
             | K::RelativeRefZero => DiagnosticKind::Resolve,
-            K::EmptyClassNegated { .. } | K::IllegalNegation { .. } => DiagnosticKind::Invalid,
+            K::EmptyClassNegated { .. } | K::IllegalNegation { .. } | K::EmptyIntersection => {
+                DiagnosticKind::Invalid
+            }
             K::CaptureInLet
             | K::ReferenceInLet
             | K::RecursiveVariable

pomsky-lib/src/exprs/char_class/char_set_item.rs

@@ -21,7 +21,7 @@ impl RegexCompoundCharSet {
         self
     }
 
-    pub(crate) fn add(mut self, other: RegexCharSet) -> Regex {
+    pub(crate) fn add(mut self, other: RegexCharSet) -> Option<Regex> {
         if other.negative && self.intersections.iter().all(|i| i.negative) {
             let mut intersections = self.intersections.into_iter();
             let mut char_set = intersections.next().expect("Intersection is empty");
@@ -32,13 +32,19 @@ impl RegexCompoundCharSet {
             if self.negative {
                 char_set = char_set.negate();
             }
-            Regex::CharSet(char_set)
-        } else {
+            Some(Regex::CharSet(char_set))
+        } else if self.may_intersect(&other) {
             self.intersections.push(other);
-            Regex::CompoundCharSet(self)
+            Some(Regex::CompoundCharSet(self))
+        } else {
+            None
         }
     }
 
+    fn may_intersect(&self, other: &RegexCharSet) -> bool {
+        self.intersections.iter().any(|set| set.may_intersect(other))
+    }
+
     pub(crate) fn codegen(&self, buf: &mut String, flavor: RegexFlavor) {
         if self.negative {
             buf.push_str("[^");
@@ -76,6 +82,10 @@ impl RegexCharSet {
         self
     }
 
+    pub(crate) fn may_intersect(&self, other: &Self) -> bool {
+        self.negative || other.negative || self.set.may_intersect(&other.set)
+    }
+
     pub(crate) fn codegen(&self, buf: &mut String, flavor: RegexFlavor, inside_compound: bool) {
         if self.set.len() == 1 {
             if let Some(range) = self.set.ranges().next() {

pomsky-lib/src/exprs/intersection.rs

@@ -24,7 +24,9 @@ impl Compile for Intersection {
         let (first_span, first) = rules.next().expect("Intersection is empty");
 
         let regex = rules.try_fold(first?, |a, (right_span, b)| match as_sets(a, b?) {
-            Ok((left, right)) => Ok(left.add(right)),
+            Ok((left, right)) => left
+                .add(right)
+                .ok_or_else(|| CompileErrorKind::EmptyIntersection.at(first_span.join(right_span))),
             Err(kind) => Err(kind.at(first_span.join(right_span))),
         })?;
 

pomsky-lib/src/unicode_set.rs

@@ -177,6 +177,12 @@ impl UnicodeSet {
             self.add(range);
         }
     }
+
+    pub(crate) fn may_intersect(&self, other: &UnicodeSet) -> bool {
+        !self.props.is_empty()
+            || !other.props.is_empty()
+            || other.ranges.iter().any(|range| self.ranges.contains(range))
+    }
 }
 
 struct MaxTwoArray<T> {

pomsky-lib/tests/testcases/errors/empty_set_negated.txt

@@ -0,0 +1,6 @@
+#! expect=error
+![s !s]
+-----
+ERROR: This negated character class matches nothing
+HELP: The group is empty because it contains both `space` and `!space`, which together match every code point
+SPAN: 1..7

pomsky-lib/tests/testcases/intersections/e_distinct_strings.txt

@@ -0,0 +1,5 @@
+#! expect=error
+'a' & 'b'
+-----
+ERROR: Intersection of expressions that do not overlap
+SPAN: 0..9