Restructure matcher module (#833)

Signed-off-by: Nico Wagner <[email protected]>

src/lib.rs

@@ -7,6 +7,7 @@ pub use record::{ByteRecord, StringRecord};
 
 mod error;
 pub mod matcher;
+mod parser;
 pub mod prelude;
 pub mod primitives;
 mod record;

src/matcher.rs → src/matcher/mod.rs

@@ -1,23 +1,16 @@
-//! Various matcher against records (and record primitives).
+//! Various matcher against record primitives.
 
 pub use error::ParseMatcherError;
 pub use occurrence::OccurrenceMatcher;
 pub use operator::{BooleanOp, RelationalOp};
 pub use options::MatcherOptions;
 pub use quantifier::Quantifier;
-pub use subfield::{
-    CardinalityMatcher, ExistsMatcher, InMatcher, RegexMatcher,
-    RegexSetMatcher, RelationMatcher, SingletonMatcher,
-    SubfieldMatcher,
-};
 pub use tag::TagMatcher;
 
 mod error;
 mod occurrence;
 mod operator;
 mod options;
-mod parse;
 mod quantifier;
-mod string;
-mod subfield;
+pub mod subfield;
 mod tag;

src/matcher/occurrence.rs

@@ -1,12 +1,15 @@
+//! Matcher that can be applied on a list of [OccurrenceRef].
+
 use std::fmt::{self, Display};
 
-use winnow::Parser;
+use winnow::combinator::{alt, empty, preceded, separated_pair};
+use winnow::prelude::*;
 
-use super::parse::parse_occurrence_matcher;
 use super::ParseMatcherError;
+use crate::primitives::parse::parse_occurrence_ref;
 use crate::primitives::{Occurrence, OccurrenceRef};
 
-/// A matcher that checks for occurrences (or no occurrence).
+/// A matcher that matches against a [OccurrenceRef].
 #[derive(Debug, Clone, PartialEq)]
 pub enum OccurrenceMatcher {
     Exact(Occurrence),
@@ -97,3 +100,107 @@ impl Display for OccurrenceMatcher {
         Ok(())
     }
 }
+
+#[inline]
+fn parse_occurrence_matcher_inner(
+    i: &mut &[u8],
+) -> PResult<Occurrence> {
+    parse_occurrence_ref.map(Occurrence::from).parse_next(i)
+}
+
+#[inline]
+fn parse_occurrence_matcher_exact(
+    i: &mut &[u8],
+) -> PResult<OccurrenceMatcher> {
+    parse_occurrence_matcher_inner
+        .verify(|occurrence| occurrence.as_bytes() != b"00")
+        .map(OccurrenceMatcher::Exact)
+        .parse_next(i)
+}
+
+#[inline]
+fn parse_occurrence_matcher_range(
+    i: &mut &[u8],
+) -> PResult<OccurrenceMatcher> {
+    separated_pair(
+        parse_occurrence_matcher_inner,
+        '-',
+        parse_occurrence_matcher_inner,
+    )
+    .verify(|(min, max)| min.len() == max.len() && min < max)
+    .map(|(min, max)| OccurrenceMatcher::Range(min, max))
+    .parse_next(i)
+}
+
+pub(crate) fn parse_occurrence_matcher(
+    i: &mut &[u8],
+) -> PResult<OccurrenceMatcher> {
+    alt((
+        preceded(
+            '/',
+            alt((
+                parse_occurrence_matcher_range,
+                parse_occurrence_matcher_exact,
+                "00".value(OccurrenceMatcher::None),
+                "*".value(OccurrenceMatcher::Any),
+            )),
+        ),
+        empty.value(OccurrenceMatcher::None),
+    ))
+    .parse_next(i)
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[test]
+    fn test_parse_occurrence_matcher() -> anyhow::Result<()> {
+        macro_rules! parse_success {
+            ($i:expr, $o:expr) => {
+                let o = parse_occurrence_matcher
+                    .parse($i.as_bytes())
+                    .unwrap();
+
+                assert_eq!(o.to_string(), $i);
+                assert_eq!(o, $o);
+            };
+        }
+
+        parse_success!("", OccurrenceMatcher::None);
+        parse_success!("/*", OccurrenceMatcher::Any);
+
+        parse_success!(
+            "/01",
+            OccurrenceMatcher::Exact(Occurrence::new("01")?)
+        );
+
+        parse_success!(
+            "/999",
+            OccurrenceMatcher::Exact(Occurrence::new("999")?)
+        );
+
+        parse_success!(
+            "/01-99",
+            OccurrenceMatcher::Range(
+                Occurrence::new("01")?,
+                Occurrence::new("99")?
+            )
+        );
+
+        parse_success!(
+            "/01-99",
+            OccurrenceMatcher::Range(
+                Occurrence::new("01")?,
+                Occurrence::new("99")?
+            )
+        );
+
+        assert_eq!(
+            parse_occurrence_matcher.parse(b"/00").unwrap(),
+            OccurrenceMatcher::None
+        );
+
+        Ok(())
+    }
+}

src/matcher/operator.rs

@@ -1,5 +1,8 @@
 use std::fmt::{self, Display};
 
+use winnow::combinator::alt;
+use winnow::{PResult, Parser};
+
 /// Relational Operator
 #[derive(Debug, Clone, PartialEq, Eq)]
 pub enum RelationalOp {
@@ -130,6 +133,30 @@ impl quickcheck::Arbitrary for RelationalOp {
     }
 }
 
+/// Parse RelationalOp which can be used for string comparisons.
+#[inline]
+pub(crate) fn parse_relational_operator(
+    i: &mut &[u8],
+) -> PResult<RelationalOp> {
+    use RelationalOp::*;
+
+    alt((
+        "==".value(Equal),
+        "!=".value(NotEqual),
+        "=^".value(StartsWith),
+        "!^".value(StartsNotWith),
+        "=$".value(EndsWith),
+        "!$".value(EndsNotWith),
+        "=*".value(Similar),
+        "=?".value(Contains),
+        ">=".value(GreaterThanOrEqual),
+        ">".value(GreaterThan),
+        "<=".value(LessThanOrEqual),
+        "<".value(LessThan),
+    ))
+    .parse_next(i)
+}
+
 /// Boolean Operators.
 #[derive(Debug, Clone, PartialEq, Eq)]
 pub enum BooleanOp {
@@ -154,3 +181,37 @@ impl quickcheck::Arbitrary for BooleanOp {
         g.choose(&[Self::And, Self::Or, Self::Xor]).unwrap().clone()
     }
 }
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[test]
+    fn test_parse_relational_operator() {
+        use RelationalOp::*;
+
+        macro_rules! parse_success {
+            ($input:expr, $expected:expr) => {
+                assert_eq!(
+                    parse_relational_operator
+                        .parse($input.as_bytes())
+                        .unwrap(),
+                    $expected
+                );
+            };
+        }
+
+        parse_success!("==", Equal);
+        parse_success!("!=", NotEqual);
+        parse_success!(">=", GreaterThanOrEqual);
+        parse_success!(">", GreaterThan);
+        parse_success!("<=", LessThanOrEqual);
+        parse_success!("<", LessThan);
+        parse_success!("=^", StartsWith);
+        parse_success!("!^", StartsNotWith);
+        parse_success!("=$", EndsWith);
+        parse_success!("!$", EndsNotWith);
+        parse_success!("=*", Similar);
+        parse_success!("=?", Contains);
+    }
+}

src/matcher/quantifier.rs

@@ -1,5 +1,8 @@
 use std::fmt::{self, Display};
 
+use winnow::combinator::alt;
+use winnow::{PResult, Parser};
+
 #[derive(Debug, Clone, PartialEq, Eq, Default)]
 pub enum Quantifier {
     All,
@@ -25,3 +28,22 @@ impl quickcheck::Arbitrary for Quantifier {
         }
     }
 }
+
+#[inline]
+pub(crate) fn parse_quantifier(i: &mut &[u8]) -> PResult<Quantifier> {
+    alt(("ALL".value(Quantifier::All), "ANY".value(Quantifier::Any)))
+        .parse_next(i)
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[test]
+    fn test_parse_quantifier() {
+        use Quantifier::*;
+
+        assert_eq!(parse_quantifier.parse(b"ALL").unwrap(), All);
+        assert_eq!(parse_quantifier.parse(b"ANY").unwrap(), Any);
+    }
+}