Merge pull request #767 from epage/perf

refactor(parse): Clean up the parser
toml-rs · Jul 29, 2024 · e922b5c · e922b5c
2 parents 27b30fd + de2e737
commit e922b5c
Show file tree

Hide file tree

Showing 5 changed files with 31 additions and 46 deletions.
diff --git a/crates/toml_edit/src/parser/mod.rs b/crates/toml_edit/src/parser/mod.rs
@@ -92,56 +92,41 @@ pub(crate) mod prelude {
         winnow::Located::new(winnow::BStr::new(s))
     }
 
-    #[cfg(not(feature = "unbounded"))]
     #[derive(Copy, Clone, Debug, Default)]
     pub(crate) struct RecursionCheck {
+        #[cfg(not(feature = "unbounded"))]
         current: usize,
     }
 
     #[cfg(not(feature = "unbounded"))]
     const LIMIT: usize = 80;
 
-    #[cfg(not(feature = "unbounded"))]
     impl RecursionCheck {
-        pub(crate) fn check_depth(depth: usize) -> Result<(), super::error::CustomError> {
-            if depth < LIMIT {
-                Ok(())
-            } else {
-                Err(super::error::CustomError::RecursionLimitExceeded)
-            }
-        }
-
-        pub(crate) fn recursing(
-            mut self,
-            input: &mut Input<'_>,
-        ) -> Result<Self, winnow::error::ErrMode<ContextError>> {
-            self.current += 1;
-            if self.current < LIMIT {
-                Ok(self)
-            } else {
-                Err(winnow::error::ErrMode::from_external_error(
-                    input,
-                    winnow::error::ErrorKind::Eof,
-                    super::error::CustomError::RecursionLimitExceeded,
-                ))
+        pub(crate) fn check_depth(_depth: usize) -> Result<(), super::error::CustomError> {
+            #[cfg(not(feature = "unbounded"))]
+            if LIMIT <= _depth {
+                return Err(super::error::CustomError::RecursionLimitExceeded);
             }
-        }
-    }
 
-    #[cfg(feature = "unbounded")]
-    #[derive(Copy, Clone, Debug, Default)]
-    pub(crate) struct RecursionCheck {}
-
-    #[cfg(feature = "unbounded")]
-    impl RecursionCheck {
-        pub(crate) fn check_depth(_depth: usize) -> Result<(), super::error::CustomError> {
             Ok(())
         }
 
+        #[allow(unused_mut)]
         pub(crate) fn recursing(
-            self,
+            mut self,
             _input: &mut Input<'_>,
         ) -> Result<Self, winnow::error::ErrMode<ContextError>> {
+            #[cfg(not(feature = "unbounded"))]
+            {
+                self.current += 1;
+                if LIMIT <= self.current {
+                    return Err(winnow::error::ErrMode::from_external_error(
+                        _input,
+                        winnow::error::ErrorKind::Eof,
+                        super::error::CustomError::RecursionLimitExceeded,
+                    ));
+                }
+            }
             Ok(self)
         }
     }

diff --git a/crates/toml_edit/src/parser/numbers.rs b/crates/toml_edit/src/parser/numbers.rs
@@ -78,7 +78,7 @@ pub(crate) fn dec_int<'i>(input: &mut Input<'i>) -> PResult<&'i str> {
                 digit.void(),
             )),
         )
-            .recognize()
+            .take()
             .map(|b: &[u8]| unsafe {
                 from_utf8_unchecked(b, "`digit` and `_` filter out non-ASCII")
             })
@@ -112,7 +112,7 @@ pub(crate) fn hex_int<'i>(input: &mut Input<'i>) -> PResult<&'i str> {
                 )
                 .map(|()| ()),
             ))
-            .recognize(),
+            .take(),
         )
         .map(|b| unsafe { from_utf8_unchecked(b, "`hexdig` and `_` filter out non-ASCII") })
         .context(StrContext::Label("hexadecimal integer")),
@@ -145,7 +145,7 @@ pub(crate) fn oct_int<'i>(input: &mut Input<'i>) -> PResult<&'i str> {
                 )
                 .map(|()| ()),
             ))
-            .recognize(),
+            .take(),
         )
         .map(|b| unsafe { from_utf8_unchecked(b, "`DIGIT0_7` and `_` filter out non-ASCII") })
         .context(StrContext::Label("octal integer")),
@@ -179,7 +179,7 @@ pub(crate) fn bin_int<'i>(input: &mut Input<'i>) -> PResult<&'i str> {
                 )
                 .map(|()| ()),
             ))
-            .recognize(),
+            .take(),
         )
         .map(|b| unsafe { from_utf8_unchecked(b, "`DIGIT0_1` and `_` filter out non-ASCII") })
         .context(StrContext::Label("binary integer")),
@@ -214,7 +214,7 @@ pub(crate) fn float_<'i>(input: &mut Input<'i>) -> PResult<&'i str> {
         dec_int,
         alt((exp.void(), (frac.void(), opt(exp.void())).void())),
     )
-        .recognize()
+        .take()
         .map(|b: &[u8]| unsafe {
             from_utf8_unchecked(
                 b,
@@ -232,7 +232,7 @@ pub(crate) fn frac<'i>(input: &mut Input<'i>) -> PResult<&'i str> {
         cut_err(zero_prefixable_int)
             .context(StrContext::Expected(StrContextValue::Description("digit"))),
     )
-        .recognize()
+        .take()
         .map(|b: &[u8]| unsafe {
             from_utf8_unchecked(
                 b,
@@ -260,7 +260,7 @@ pub(crate) fn zero_prefixable_int<'i>(input: &mut Input<'i>) -> PResult<&'i str>
         )
         .map(|()| ()),
     )
-        .recognize()
+        .take()
         .map(|b: &[u8]| unsafe { from_utf8_unchecked(b, "`digit` and `_` filter out non-ASCII") })
         .parse_next(input)
 }
@@ -273,7 +273,7 @@ pub(crate) fn exp<'i>(input: &mut Input<'i>) -> PResult<&'i str> {
         opt(one_of([b'+', b'-'])),
         cut_err(zero_prefixable_int),
     )
-        .recognize()
+        .take()
         .map(|b: &[u8]| unsafe {
             from_utf8_unchecked(
                 b,

diff --git a/crates/toml_edit/src/parser/strings.rs b/crates/toml_edit/src/parser/strings.rs
@@ -321,7 +321,7 @@ fn ml_literal_body<'i>(input: &mut Input<'i>) -> PResult<&'i str> {
         .map(|()| ()),
         opt(mll_quotes(ML_LITERAL_STRING_DELIM.void())),
     )
-        .recognize()
+        .take()
         .try_map(std::str::from_utf8)
         .parse_next(input)
 }

diff --git a/crates/toml_edit/src/parser/trivia.rs b/crates/toml_edit/src/parser/trivia.rs
@@ -146,7 +146,7 @@ mod test {
         ];
         for input in inputs {
             dbg!(input);
-            let parsed = ws_comment_newline.recognize().parse(new_input(input));
+            let parsed = ws_comment_newline.take().parse(new_input(input));
             assert!(parsed.is_ok(), "{:?}", parsed);
             let parsed = parsed.unwrap();
             assert_eq!(parsed, input.as_bytes());

diff --git a/crates/toml_edit/src/parser/value.rs b/crates/toml_edit/src/parser/value.rs
@@ -81,12 +81,12 @@ pub(crate) fn value<'i>(check: RecursionCheck) -> impl Parser<Input<'i>, Value,
             },
     }
         .with_span()
-        .try_map(|(value, span)| apply_raw(value, span))
+        .map(|(value, span)| apply_raw(value, span))
         .parse_next(input)
     }
 }
 
-fn apply_raw(mut val: Value, span: std::ops::Range<usize>) -> Result<Value, std::str::Utf8Error> {
+fn apply_raw(mut val: Value, span: std::ops::Range<usize>) -> Value {
     match val {
         Value::String(ref mut f) => {
             let raw = RawString::with_span(span);
@@ -116,7 +116,7 @@ fn apply_raw(mut val: Value, span: std::ops::Range<usize>) -> Result<Value, std:
         }
     };
     val.decorate("", "");
-    Ok(val)
+    val
 }
 
 #[cfg(test)]