Make StringOrInit use COW for string variant to avoid a copy when calling

quirks::process_match_input.

Author: edgul

benches/parse_patterns.rs

@@ -6,7 +6,7 @@ fn bench_parse_shipping_groups_summary(c: &mut Criterion) {
         b.iter(|| {
             let input = quirks::process_construct_pattern_input(
                 black_box(quirks::StringOrInit::String(
-                    "component-ShippingGroupsSummary.*.js".to_owned()
+                    "component-ShippingGroupsSummary.*.js".into()
                 )),
                 black_box(Some("https://example.test/web/")),
             );

src/lib.rs

@@ -659,10 +659,11 @@ mod tests {
 
   #[derive(Debug, Deserialize)]
   #[serde(untagged)]
+  #[serde(bound(deserialize = "'de: 'a"))]
   #[allow(clippy::large_enum_variant)]
-  enum ExpectedMatch {
+  enum ExpectedMatch<'a> {
     String(String),
-    MatchResult(MatchResult),
+    MatchResult(MatchResult<'a>),
   }
 
   #[derive(Debug, Deserialize)]
@@ -674,28 +675,30 @@ mod tests {
   #[allow(clippy::large_enum_variant)]
   #[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize)]
   #[serde(untagged)]
-  pub enum StringOrInitOrOptions {
+  pub enum StringOrInitOrOptions<'a> {
     Options(UrlPatternOptions),
-    StringOrInit(quirks::StringOrInit),
+    StringOrInit(quirks::StringOrInit<'a>),
   }
 
   #[derive(Debug, Deserialize)]
-  struct TestCase {
+  #[serde(bound(deserialize = "'de: 'a"))]
+  struct TestCase<'a> {
     skip: Option<String>,
-    pattern: Vec<StringOrInitOrOptions>,
+    pattern: Vec<StringOrInitOrOptions<'a>>,
     #[serde(default)]
-    inputs: Vec<quirks::StringOrInit>,
-    expected_obj: Option<quirks::StringOrInit>,
-    expected_match: Option<ExpectedMatch>,
+    inputs: Vec<quirks::StringOrInit<'a>>,
+    expected_obj: Option<quirks::StringOrInit<'a>>,
+    expected_match: Option<ExpectedMatch<'a>>,
     #[serde(default)]
     exactly_empty_components: Vec<String>,
   }
 
   #[derive(Debug, Deserialize)]
-  struct MatchResult {
+  #[serde(bound(deserialize = "'de: 'a"))]
+  struct MatchResult<'a> {
     #[serde(deserialize_with = "deserialize_match_result_inputs")]
     #[serde(default)]
-    inputs: Option<(quirks::StringOrInit, Option<String>)>,
+    inputs: Option<(quirks::StringOrInit<'a>, Option<String>)>,
 
     protocol: Option<ComponentResult>,
     username: Option<ComponentResult>,
@@ -707,17 +710,17 @@ mod tests {
     hash: Option<ComponentResult>,
   }
 
-  fn deserialize_match_result_inputs<'de, D>(
+  fn deserialize_match_result_inputs<'a, D>(
     deserializer: D,
-  ) -> Result<Option<(quirks::StringOrInit, Option<String>)>, D::Error>
+  ) -> Result<Option<(quirks::StringOrInit<'a>, Option<String>)>, D::Error>
   where
-    D: serde::Deserializer<'de>,
+    D: serde::Deserializer<'a>,
   {
     #[derive(Debug, Deserialize)]
     #[serde(untagged)]
-    enum MatchResultInputs {
-      OneArgument((quirks::StringOrInit,)),
-      TwoArguments(quirks::StringOrInit, String),
+    enum MatchResultInputs<'a> {
+      OneArgument((quirks::StringOrInit<'a>,)),
+      TwoArguments(quirks::StringOrInit<'a>, String),
     }
 
     let res = Option::<MatchResultInputs>::deserialize(deserializer)?;
@@ -811,7 +814,7 @@ mod tests {
       ..
     }) = &input
     {
-      base_url = Some(url.clone())
+      base_url = Some(url.clone().into())
     }
 
     macro_rules! assert_field {
@@ -921,7 +924,8 @@ mod tests {
 
     let input = input.unwrap_or_else(|| StringOrInit::Init(Default::default()));
 
-    let expected_input = (input.clone(), base_url.clone());
+    let expected_input =
+      (input.clone(), base_url.clone().map(|s| s.to_string()));
 
     let match_input = quirks::process_match_input(input, base_url.as_deref());
 
@@ -1058,7 +1062,7 @@ mod tests {
   #[test]
   fn issue46() {
     quirks::process_construct_pattern_input(
-      quirks::StringOrInit::String(":café://:foo".to_owned()),
+      quirks::StringOrInit::String(":café://:foo".to_owned().into()),
       None,
     )
     .unwrap();

src/quirks.rs

@@ -3,6 +3,7 @@
 
 use serde::Deserialize;
 use serde::Serialize;
+use std::borrow::Cow;
 use url::Url;
 
 pub use crate::Error;
@@ -36,8 +37,8 @@ pub struct UrlPatternInit {
 #[allow(clippy::large_enum_variant)]
 #[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize)]
 #[serde(untagged)]
-pub enum StringOrInit {
-  String(String),
+pub enum StringOrInit<'a> {
+  String(Cow<'a, str>),
   Init(UrlPatternInit),
 }
 
@@ -106,7 +107,7 @@ impl<R: RegExp> From<Component<R>> for UrlPatternComponent {
   fn from(component: Component<R>) -> Self {
     let regexp_string = component
       .regexp
-      .map(|r| r.pattern_string())
+      .map(|r| r.pattern_string().to_owned())
       .unwrap_or_default();
     Self {
       pattern_string: component.pattern_string,
@@ -166,7 +167,9 @@ impl<R: RegExp> From<crate::matcher::InnerMatcher<R>> for InnerMatcher {
         allow_empty,
       },
       crate::matcher::InnerMatcher::RegExp { regexp } => Self::RegExp {
-        regexp: regexp.map(|r| r.pattern_string()).unwrap_or_default(),
+        regexp: regexp
+          .map(|r| r.pattern_string().to_owned())
+          .unwrap_or_default(),
       },
     }
   }
@@ -196,8 +199,8 @@ impl RegExp for EcmaRegexp {
     regexp.matches(text)
   }
 
-  fn pattern_string(&self) -> String {
-    self.0.clone()
+  fn pattern_string(&self) -> &str {
+    self.0.as_ref()
   }
 }
 
@@ -230,12 +233,12 @@ pub fn parse_pattern_as_lib<R: RegExp>(
   Ok(pattern)
 }
 
-pub type Inputs = (StringOrInit, Option<String>);
+pub type Inputs<'a> = (StringOrInit<'a>, Option<String>);
 
-pub fn process_match_input(
-  input: StringOrInit,
+pub fn process_match_input<'a>(
+  input: StringOrInit<'a>,
   base_url_str: Option<&str>,
-) -> Result<Option<(crate::UrlPatternMatchInput, Inputs)>, Error> {
+) -> Result<Option<(crate::UrlPatternMatchInput, Inputs<'a>)>, Error> {
   let mut inputs = (input.clone(), None);
   let init = match input {
     StringOrInit::String(url) => {

src/regexp.rs

@@ -17,7 +17,7 @@ pub trait RegExp: Sized {
   /// Returns `None` if the text does not match the regular expression.
   fn matches<'a>(&self, text: &'a str) -> Option<Vec<Option<&'a str>>>;
 
-  fn pattern_string(&self) -> String;
+  fn pattern_string(&self) -> &str;
 }
 
 impl RegExp for regex::Regex {
@@ -41,7 +41,7 @@ impl RegExp for regex::Regex {
     Some(captures)
   }
 
-  fn pattern_string(&self) -> String {
-    self.as_str().to_string()
+  fn pattern_string(&self) -> &str {
+    self.as_str()
   }
 }