optimize regex and string allocation

Author: Elias Ram

Cargo.lock

@@ -32,15 +32,14 @@ UNSTABLE_cadence = ["dep:cadence", "UNSTABLE_metrics"]
 [dependencies]
 cadence = { version = "0.29.0", optional = true }
 crc32fast = "1.4.0"
-itertools = "0.10.5"
+itertools = "0.13.0"
 log = { version = "0.4.8", optional = true, features = ["std"] }
 once_cell = "1"
 rand = { version = "0.8.1", optional = true }
 regex = "1.7.3"
 sentry-types = { version = "0.32.3", path = "../sentry-types" }
 serde = { version = "1.0.104", features = ["derive"] }
 serde_json = { version = "1.0.46" }
-unicode-segmentation = "1.11.0"
 uuid = { version = "1.0.0", features = ["v4", "serde"], optional = true }
 
 [dev-dependencies]

sentry-core/Cargo.toml

@@ -535,13 +535,13 @@ impl Metric {
         let data = format!(
             "{}@{}:{}|{}|#{}|T{}",
             NormalizedName::from(self.name.as_ref()),
-            NormalizedUnit::from(self.unit),
+            NormalizedUnit::from(self.unit.to_string().as_ref()),
             self.value,
             self.value.ty(),
-            NormalizedTags::from(self.tags),
+            NormalizedTags::from(&self.tags),
             timestamp
         );
-        Envelope::from_item(EnvelopeItem::Statsd(data.into_bytes()))
+        EnvelopeItem::Statsd(data.into_bytes()).into()
     }
 }
 
@@ -597,9 +597,9 @@ impl MetricBuilder {
         K: Into<MetricStr>,
         V: Into<MetricStr>,
     {
-        tags.into_iter().for_each(|(k, v)| {
+        for (k, v) in tags {
             self.metric.tags.insert(k.into(), v.into());
-        });
+        }
         self
     }
 
@@ -836,7 +836,12 @@ impl Worker {
         for (timestamp, buckets) in buckets {
             for (key, value) in buckets {
                 write!(&mut out, "{}", NormalizedName::from(key.name.as_ref()))?;
-                write!(&mut out, "@{}", NormalizedUnit::from(key.unit))?;
+                match key.unit {
+                    MetricUnit::Custom(u) => {
+                        write!(&mut out, "@{}", NormalizedUnit::from(u.as_ref()))?
+                    }
+                    _ => write!(&mut out, "@{}", key.unit)?,
+                }
                 match value {
                     BucketValue::Counter(c) => {
                         write!(&mut out, ":{}", c)?;
@@ -862,7 +867,7 @@ impl Worker {
 
                 write!(&mut out, "|{}", key.ty.as_str())?;
                 let normalized_tags =
-                    NormalizedTags::from(key.tags).with_default_tags(&self.default_tags);
+                    NormalizedTags::from(&key.tags).with_default_tags(&self.default_tags);
                 write!(&mut out, "|#{}", normalized_tags)?;
                 writeln!(&mut out, "|T{}", timestamp)?;
             }

sentry-core/src/metrics/mod.rs

@@ -1,3 +1,24 @@
 pub mod normalized_name;
 pub mod normalized_tags;
 pub mod normalized_unit;
+
+pub fn truncate(s: &str, max_chars: usize) -> &str {
+    match s.char_indices().nth(max_chars) {
+        None => s,
+        Some((i, _)) => &s[..i],
+    }
+}
+
+#[cfg(test)]
+mod test {
+
+    #[test]
+    fn test_truncate_ascii_chars() {
+        assert_eq!("abc", super::truncate("abcde", 3));
+    }
+
+    #[test]
+    fn test_truncate_unicode_chars() {
+        assert_eq!("😀😀😀", super::truncate("😀😀😀😀😀", 3));
+    }
+}

sentry-core/src/metrics/normalization/mod.rs

@@ -1,22 +1,24 @@
+use std::{borrow::Cow, sync::OnceLock};
+
 use regex::Regex;
-use std::borrow::Cow;
 
 pub struct NormalizedName<'a> {
     name: Cow<'a, str>,
 }
 
 impl<'a> From<&'a str> for NormalizedName<'a> {
     fn from(name: &'a str) -> Self {
+        static METRIC_NAME_RE: OnceLock<Regex> = OnceLock::new();
         Self {
-            name: Regex::new(r"[^a-zA-Z0-9_\-.]")
-                .expect("Regex should compile")
-                .replace_all(name, "_"),
+            name: METRIC_NAME_RE
+                .get_or_init(|| Regex::new(r"[^a-zA-Z0-9_\-.]").expect("Regex should compile"))
+                .replace_all(super::truncate(name, 150), "_"),
         }
     }
 }
 
 impl std::fmt::Display for NormalizedName<'_> {
-    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
+    fn fmt(&self, f: &mut std::fmt::Formatter) -> std::fmt::Result {
         write!(f, "{}", self.name)
     }
 }
@@ -33,4 +35,13 @@ mod test {
 
         assert_eq!(expected, actual);
     }
+
+    #[test]
+    fn test_length_restriction() {
+        let expected = "a".repeat(150);
+
+        let actual = NormalizedName::from("a".repeat(155).as_ref()).to_string();
+
+        assert_eq!(expected, actual);
+    }
 }