fix(perf): Improve performance of event processing by avoiding regex clone

CHANGELOG.md

@@ -25,6 +25,7 @@
 - Add the client SDK to session kafka payloads. ([#751](https://github.com/getsentry/relay/pull/751))
 - Add a standalone tool to document metrics in JSON or YAML. ([#752](https://github.com/getsentry/relay/pull/752))
 - Emit `processing.event.produced` for user report and session Kafka messages. ([#757](https://github.com/getsentry/relay/pull/757))
+- Improve performance of event processing by avoiding regex clone. ([#767](https://github.com/getsentry/relay/pull/767))
 
 ## 20.8.0
 

relay-general/derive/src/lib.rs

@@ -728,12 +728,9 @@ impl FieldAttrs {
         };
 
         let match_regex = if let Some(ref match_regex) = self.match_regex {
-            quote!(Some(
-                #[allow(clippy::trivial_regex)]
-                ::regex::Regex::new(#match_regex).unwrap()
-            ))
+            quote!(Some(#match_regex))
         } else if let Some(ref parent_attrs) = inherit_from_field_attrs {
-            quote!(#parent_attrs.match_regex.clone())
+            quote!(#parent_attrs.match_regex)
         } else {
             quote!(None)
         };

relay-general/derive/src/process.rs

@@ -81,10 +81,7 @@ pub fn derive_process_value(mut s: synstructure::Structure<'_>) -> TokenStream {
             let field_attrs_tokens = field_attrs.as_tokens(None);
 
             (quote! {
-                ::lazy_static::lazy_static! {
-                    static ref #field_attrs_name: crate::processor::FieldAttrs =
-                        #field_attrs_tokens;
-                }
+                static #field_attrs_name: crate::processor::FieldAttrs = #field_attrs_tokens;
             })
             .to_tokens(&mut body);
 
@@ -94,21 +91,21 @@ pub fn derive_process_value(mut s: synstructure::Structure<'_>) -> TokenStream {
                 }
 
                 quote! {
-                    __state.enter_nothing(Some(::std::borrow::Cow::Borrowed(&*#field_attrs_name)))
+                    __state.enter_nothing(Some(::std::borrow::Cow::Borrowed(&#field_attrs_name)))
                 }
             } else if is_tuple_struct {
                 quote! {
                     __state.enter_index(
                         #index,
-                        Some(::std::borrow::Cow::Borrowed(&*#field_attrs_name)),
+                        Some(::std::borrow::Cow::Borrowed(&#field_attrs_name)),
                         crate::processor::ValueType::for_field(#ident),
                     )
                 }
             } else {
                 quote! {
                     __state.enter_static(
                         #field_name,
-                        Some(::std::borrow::Cow::Borrowed(&*#field_attrs_name)),
+                        Some(::std::borrow::Cow::Borrowed(&#field_attrs_name)),
                         crate::processor::ValueType::for_field(#ident),
                     )
                 }

relay-general/src/processor/attrs.rs

@@ -3,7 +3,6 @@ use std::fmt;
 use std::str::FromStr;
 
 use failure::Fail;
-use regex::Regex;
 use smallvec::SmallVec;
 
 use crate::processor::{ProcessValue, SelectorPathItem, SelectorSpec};
@@ -232,7 +231,7 @@ pub enum Pii {
 }
 
 /// Meta information about a field.
-#[derive(Debug, Clone)]
+#[derive(Debug, Clone, Copy)]
 pub struct FieldAttrs {
     /// Optionally the name of the field.
     pub name: Option<&'static str>,
@@ -243,7 +242,7 @@ pub struct FieldAttrs {
     /// Whether to trim whitespace from this string.
     pub trim_whitespace: bool,
     /// A regex to validate the (string) value against.
-    pub match_regex: Option<Regex>,
+    pub match_regex: Option<&'static str>,
     /// The maximum char length of this field.
     pub max_chars: Option<MaxChars>,
     /// The maximum bag size of this field.

relay-general/src/store/schema.rs

@@ -1,10 +1,40 @@
+use regex::Regex;
+
 use crate::processor::{ProcessValue, ProcessingState, Processor};
 use crate::types::{
     Array, Empty, Error, ErrorKind, Meta, Object, ProcessingAction, ProcessingResult,
 };
 
 pub struct SchemaProcessor;
 
+macro_rules! declare_used_field_regexes {
+    ($($ident:ident: $regex:expr),* $(,)*) => {
+        fn get_regex(name: &'static str) -> &'static Regex {
+            lazy_static::lazy_static! {
+                $(
+                    static ref $ident: Regex = Regex::new($regex).unwrap();
+                )*
+            };
+
+            match name {
+                $($regex => &$ident, )*
+                _ => panic!("Please declare Regex {} using declare_used_field_regexes.", name),
+            }
+        }
+    }
+}
+
+// Pre-built list of regexes for max performance. The identifier in front is arbitrary, but needs
+// to be unique.
+declare_used_field_regexes![
+    A: r"^[^\r\n\f\t/]*\z",
+    B: r"^[^\r\n\x0C/]+$",
+    C: r"^[^\r\n]*\z",
+    D: r"^[a-zA-Z0-9_\.:-]+\z",
+    E: r"^\s*[a-zA-Z0-9_.-]*\s*$",
+    F: r"^[^\n]+\z",
+];
+
 impl Processor for SchemaProcessor {
     fn process_string(
         &mut self,
@@ -95,8 +125,8 @@ fn verify_value_pattern(
     meta: &mut Meta,
     state: &ProcessingState<'_>,
 ) -> ProcessingResult {
-    if let Some(ref regex) = state.attrs().match_regex {
-        if !regex.is_match(value) {
+    if let Some(ref regex_string) = state.attrs().match_regex {
+        if !get_regex(regex_string).is_match(value) {
             meta.add_error(Error::invalid("invalid characters in string"));
             return Err(ProcessingAction::DeleteValueSoft);
         }