fix: speed up Position::line_col for large inputs using SIMD

derive/Cargo.toml

@@ -1,7 +1,7 @@
 [package]
 name = "pest_derive"
 description = "pest's derive macro"
-version = "2.3.0"
+version = "2.3.1"
 edition = "2018"
 authors = ["Dragoș Tiselice <dragostiselice@gmail.com>"]
 homepage = "https://pest-parser.github.io/"
@@ -23,5 +23,5 @@ std = ["pest/std", "pest_generator/std"]
 
 [dependencies]
 # for tests, included transitively anyway
-pest = { path = "../pest", version = "2.3.0", default-features = false }
-pest_generator = { path = "../generator", version = "2.3.0", default-features = false }
+pest = { path = "../pest", version = "2.3.1", default-features = false }
+pest_generator = { path = "../generator", version = "2.3.1", default-features = false }

generator/Cargo.toml

@@ -1,7 +1,7 @@
 [package]
 name = "pest_generator"
 description = "pest code generator"
-version = "2.3.0"
+version = "2.3.1"
 edition = "2018"
 authors = ["Dragoș Tiselice <dragostiselice@gmail.com>"]
 homepage = "https://pest-parser.github.io/"
@@ -18,8 +18,8 @@ default = ["std"]
 std = ["pest/std"]
 
 [dependencies]
-pest = { path = "../pest", version = "2.3.0", default-features = false }
-pest_meta = { path = "../meta", version = "2.3.0" }
+pest = { path = "../pest", version = "2.3.1", default-features = false }
+pest_meta = { path = "../meta", version = "2.3.1" }
 proc-macro2 = "1.0"
 quote = "1.0"
 syn = "1.0"

grammars/Cargo.toml

@@ -1,7 +1,7 @@
 [package]
 name = "pest_grammars"
 description = "pest popular grammar implementations"
-version = "2.3.0"
+version = "2.3.1"
 edition = "2018"
 authors = ["Dragoș Tiselice <dragostiselice@gmail.com>"]
 homepage = "https://pest-parser.github.io/"
@@ -14,8 +14,8 @@ readme = "_README.md"
 rust-version = "1.56"
 
 [dependencies]
-pest = { path = "../pest", version = "2.3.0" }
-pest_derive = { path = "../derive", version = "2.3.0" }
+pest = { path = "../pest", version = "2.3.1" }
+pest_derive = { path = "../derive", version = "2.3.1" }
 
 [dev-dependencies]
 criterion = "0.3"

grammars/benches/json.rs

@@ -30,5 +30,45 @@ fn criterion_benchmark(c: &mut Criterion) {
     });
 }
 
-criterion_group!(benches, criterion_benchmark);
+mod autocorrect {
+    use pest_derive::Parser;
+
+    #[derive(Parser)]
+    #[grammar_inline = r#"
+newline = ${ "\n" | "\r" }
+space = ${ " "+ }
+
+other = ${ !(pair) ~ ANY }
+comment = ${ single_line_comment | multiline_comment }
+single_line_comment = _{ "//" ~ (!(newline) ~ ANY)* }
+multiline_comment = _{ "/*" ~ (!("*/") ~ ANY)* ~ "*/"}
+
+string_type = _{ 
+  ("\"" ~ (!(newline | "\"") ~ ANY)* ~ "\"") 
+}
+key = ${ string_type ~ (" ")* ~ ":" ~ (" ")* }
+string = ${ string_type  }
+pair = _{ key ~ string }
+
+line = _{ pair | comment | space | other | newline }
+item = _{ SOI ~ line* ~ EOI } 
+"#]
+    pub struct JsonParser;
+}
+
+fn line_col_benchmark(c: &mut Criterion) {
+    let mut file = File::open("benches/main.i18n.json").unwrap();
+    let mut data = String::new();
+
+    file.read_to_string(&mut data).unwrap();
+    let pairs = autocorrect::JsonParser::parse(autocorrect::Rule::item, &data).unwrap();
+    let last_pair = pairs.last().unwrap();
+    c.bench_function("line col", |b| {
+        b.iter(|| {
+            let _ = last_pair.as_span().start_pos().line_col();
+        });
+    });
+}
+
+criterion_group!(benches, criterion_benchmark, line_col_benchmark,);
 criterion_main!(benches);

meta/Cargo.toml

@@ -1,7 +1,7 @@
 [package]
 name = "pest_meta"
 description = "pest meta language parser and validator"
-version = "2.3.0"
+version = "2.3.1"
 edition = "2018"
 authors = ["Dragoș Tiselice <dragostiselice@gmail.com>"]
 homepage = "https://pest-parser.github.io/"
@@ -16,7 +16,7 @@ include = ["Cargo.toml", "src/**/*", "src/grammar.rs", "_README.md", "LICENSE-*"
 rust-version = "1.56"
 
 [dependencies]
-pest = { path = "../pest", version = "2.3.0" }
+pest = { path = "../pest", version = "2.3.1" }
 once_cell = "1.8.0"
 
 [build-dependencies]

pest/Cargo.toml

@@ -1,7 +1,7 @@
 [package]
 name = "pest"
 description = "The Elegant Parser"
-version = "2.3.0"
+version = "2.3.1"
 edition = "2018"
 authors = ["Dragoș Tiselice <dragostiselice@gmail.com>"]
 homepage = "https://pest-parser.github.io/"
@@ -21,9 +21,14 @@ std = ["ucd-trie/std", "thiserror"]
 pretty-print = ["serde", "serde_json"]
 # Enable const fn constructor for `PrecClimber`
 const_prec_climber = []
+# Enable faster `Position::line_col` calculation using SIMD
+# (note that this may have extra overhead for small inputs)
+fast-line-col = ["memchr", "bytecount"]
 
 [dependencies]
 ucd-trie = { version = "0.1.1", default-features = false }
 serde = { version = "1.0.89", optional = true }
 serde_json = { version = "1.0.39", optional = true}
 thiserror = { version = "1.0.31", optional = true }
+memchr = { version = "2", optional = true }
+bytecount = { version = "0.6", optional = true }

pest/src/position.rs

@@ -135,45 +135,14 @@ impl<'i> Position<'i> {
         if self.pos > self.input.len() {
             panic!("position out of bounds");
         }
-
-        let mut pos = self.pos;
-        // Position's pos is always a UTF-8 border.
-        let slice = &self.input[..pos];
-        let mut chars = slice.chars().peekable();
-
-        let mut line_col = (1, 1);
-
-        while pos != 0 {
-            match chars.next() {
-                Some('\r') => {
-                    if let Some(&'\n') = chars.peek() {
-                        chars.next();
-
-                        if pos == 1 {
-                            pos -= 1;
-                        } else {
-                            pos -= 2;
-                        }
-
-                        line_col = (line_col.0 + 1, 1);
-                    } else {
-                        pos -= 1;
-                        line_col = (line_col.0, line_col.1 + 1);
-                    }
-                }
-                Some('\n') => {
-                    pos -= 1;
-                    line_col = (line_col.0 + 1, 1);
-                }
-                Some(c) => {
-                    pos -= c.len_utf8();
-                    line_col = (line_col.0, line_col.1 + 1);
-                }
-                None => unreachable!(),
-            }
+        #[cfg(feature = "fast-line-col")]
+        {
+            fast_line_col(self.input, self.pos)
+        }
+        #[cfg(not(feature = "fast-line-col"))]
+        {
+            original_line_col(self.input, self.pos)
         }
-
-        line_col
     }
 
     /// Returns the entire line of the input that contains this `Position`.
@@ -432,6 +401,63 @@ impl<'i> Hash for Position<'i> {
     }
 }
 
+#[inline]
+#[cfg(not(feature = "fast-line-col"))]
+fn original_line_col(input: &str, mut pos: usize) -> (usize, usize) {
+    // Position's pos is always a UTF-8 border.
+    let slice = &input[..pos];
+    let mut chars = slice.chars().peekable();
+
+    let mut line_col = (1, 1);
+
+    while pos != 0 {
+        match chars.next() {
+            Some('\r') => {
+                if let Some(&'\n') = chars.peek() {
+                    chars.next();
+
+                    if pos == 1 {
+                        pos -= 1;
+                    } else {
+                        pos -= 2;
+                    }
+
+                    line_col = (line_col.0 + 1, 1);
+                } else {
+                    pos -= 1;
+                    line_col = (line_col.0, line_col.1 + 1);
+                }
+            }
+            Some('\n') => {
+                pos -= 1;
+                line_col = (line_col.0 + 1, 1);
+            }
+            Some(c) => {
+                pos -= c.len_utf8();
+                line_col = (line_col.0, line_col.1 + 1);
+            }
+            None => unreachable!(),
+        }
+    }
+
+    line_col
+}
+
+#[inline]
+#[cfg(feature = "fast-line-col")]
+fn fast_line_col(input: &str, pos: usize) -> (usize, usize) {
+    // Position's pos is always a UTF-8 border.
+    let slice = &input[..pos];
+
+    let prec_ln = memchr::memrchr(b'\n', slice.as_bytes());
+    if let Some(prec_nl_pos) = prec_ln {
+        let lines = bytecount::count(slice[..=prec_nl_pos].as_bytes(), b'\n') + 1;
+        (lines, slice[prec_nl_pos..].chars().count())
+    } else {
+        (1, slice.chars().count() + 1)
+    }
+}
+
 #[cfg(test)]
 mod tests {
     use super::*;
@@ -465,6 +491,8 @@ mod tests {
         assert_eq!(Position::new(input, 7).unwrap().line_col(), (3, 1));
         assert_eq!(Position::new(input, 8).unwrap().line_col(), (3, 2));
         assert_eq!(Position::new(input, 11).unwrap().line_col(), (3, 3));
+        let input = "abcd嗨";
+        assert_eq!(Position::new(input, 7).unwrap().line_col(), (1, 6));
     }
 
     #[test]

vm/Cargo.toml

@@ -1,7 +1,7 @@
 [package]
 name = "pest_vm"
 description = "pest grammar virtual machine"
-version = "2.3.0"
+version = "2.3.1"
 edition = "2018"
 authors = ["Dragoș Tiselice <dragostiselice@gmail.com>"]
 homepage = "https://pest-parser.github.io/"
@@ -14,5 +14,5 @@ readme = "_README.md"
 rust-version = "1.56"
 
 [dependencies]
-pest = { path = "../pest", version = "2.3.0" }
-pest_meta = { path = "../meta", version = "2.3.0" }
+pest = { path = "../pest", version = "2.3.1" }
+pest_meta = { path = "../meta", version = "2.3.1" }